(Advances in agronomy 114) donald l sparks (eds ) advances in agronomy 114 academic press, elsevier (2012)

Thus, inaddition to being able to source germplasm from mega-environmentswith conditions similar to those arising from climate change in their ownareas, breeders will need the capacity t

Texas A&M University

Emeritus Advisory Board Members

Prepared in cooperation with the

American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America Book and Multimedia Publishing Committee

DAVID D BALTENSPERGER, CHAIR

SALLY D LOGSDON

225 Wyman Street, Waltham, MA02451, USA

32 Jamestown Road, London, NW1 7BY, UK

Radarweg 29, PO Box 211, 1000 AE Amsterdam, The Netherlands

First edition 2012

Copyright # 2012 Elsevier Inc All rights reserved.

No part of this publication may be reproduced, stored in a retrieval system or transmitted

in any form or by any means electronic, mechanical, photocopying, recording or otherwise without the prior written permission of the publisher

Permissions may be sought directly from Elsevier’s Science & Technology Rights Department in Oxford, UK: phone (+44) (0) 1865 843830; fax (+44) (0) 1865 853333; email: permissions@elsevier.com Alternatively you can submit your request online by visiting the Elsevier web site at http://elsevier.com/locate/permissions, and selecting Obtaining permission to use Elsevier material

Notice

No responsibility is assumed by the publisher for any injury and/or damage to persons or property as a matter of products liability, negligence or otherwise, or from any use or operation of any methods, products, instructions or ideas contained in the material herein Because of rapid advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made

ISBN: 978-0-12-394275-3

ISSN: 0065-2113 (series)

For information on all Academic Press publications

visit our website at elsevierdirect.com

Printed and bound in USA

12 13 14 15 10 9 8 7 6 5 4 3 2 1

Numbers in Parentheses indicate the pages on which the authors’ contributions begin.

Departamento de Ciencias Agroforestales, ETSIA Universidad de Sevilla, Sevilla, Spain

Environmental Science Program and Quesnel River Research Centre, University

of Northern British Columbia, Prince George, British Columbia, Canada

Allison Rick VandeVoort (59)

School of Agricultural, Forest, and Environmental Sciences, Clemson University, Clemson, South Carolina, USA

Departamento de Agronomı´a, Universidad de Co´rdoba, Co´rdoba, Spain

Eldert J van Henten (155)

Farm Technology Group, Wageningen University, and Wageningen UR house Horticulture, Wageningen, The Netherlands

Green-N Verhulst (1)

International Maize and Wheat Improvement Centre (CIMMYT), Mexico D.F., Mexico, and Department of Earth and Environmental Sciences, Katholieke Universiteit Leuven, Leuven, Belgium

Volume 114 of Advances in Agronomy contains six excellent and timelyreviews dealing with plant, soil, and environmental sciences Chapter 1 is

a review on adaptation and mitigation strategies for producing maize in achanging climate Emphasis is placed on advances in stress tolerance breed-ing and physiology to develop rapid germplasm for a changing environ-ment Chapter 2 is a comprehensive overview on the environmentalchemistry of silver in soils In addition to discussion on the geochemistry

of silver, coverage is provided on silver nanoparticle technology and thereactivity of silver nanoparticles in the soil environment Chapter 3 discussesthe important role that phosphorus plays in agriculture and the environment

in West Asia and North Africa Chapter 4 is an interesting overview onways to sense soil properties in situ and online in the laboratory Differenttypes of sensors and their applications are discussed Chapter 5 presents aprototype decision support system for effective design and placement ofvegetated buffer strips in field situations to mitigate sediment transport anddeposition Chapter 6 is a review on biological nitrification inhibitionstrategies in agricultural settings and effects on the global environment

I appreciate the fine contributions of the authors

DONALDL SPARKSNewark, Delaware, USA

xiii

Maize Production in a Changing

Climate: Impacts, Adaptation, and

Mitigation Strategies

J E Cairns,*K Sonder,*P H Zaidi,†N Verhulst,*,‡G Mahuku,*

R Babu,*S K Nair,*B Das,§B Govaerts,*M T Vinayan,†

Z Rashid,†J J Noor,†P Devi,†F San Vicente,*and

B M Prasanna§

Contents

2 Likely Climate Scenarios for Sub-Saharan Africa and South Asia

3 Adaptation Technologies and Practices for Addressing Near-Term

3.1 Abiotic stresses—Drought, heat, and waterlogging 11 3.2 Biotic stresses of maize under the changing climate 20 3.3 Strategies for mitigating climate-related effects of

3.4 Breeding approaches for tolerance to climate-related stresses 24 3.5 Crop management options for increasing the resilience

4 Mitigation Technologies and Practices for Reducing Greenhouse

4.2 Management practices to reduce the global warming potential

* International Maize and Wheat Improvement Centre (CIMMYT), Mexico D.F., Mexico

{ International Maize and Wheat Improvement Centre (CIMMYT), Hyderabad, India

{ Department of Earth and Environmental Sciences, Katholieke Universiteit Leuven, Leuven, Belgium} International Maize and Wheat Improvement Centre (CIMMYT), Nairobi, Kenya

1

Plant breeding and improved management options have made remarkable progress in increasing crop yields during the past century However, climate change projections suggest that large yield losses will be occurring in many regions, particularly within sub-Saharan Africa The development of climate- ready germplasm to offset these losses is of the upmost importance Given the time lag between the development of improved germplasm and adoption in farmers’ fields, the development of improved breeding pipelines needs to be a high priority Recent advances in molecular breeding provide powerful tools to accelerate breeding gains and dissect stress adaptation This review focuses on achievements in stress tolerance breeding and physiology and presents future tools for quick and efficient germplasm development Sustainable agronomic and resource management practices can effectively contribute to climate change mitigation Management options to increase maize system resilience

to climate-related stresses and mitigate the effects of future climate change are also discussed.

1 Introduction

Maize is produced on nearly 100 million hectares in developingcountries, with almost 70% of the total maize production in the developingworld coming from low and lower middle income countries (FAOSTAT,

2010) By 2050, demand for maize will double in the developing world, andmaize is predicted to become the crop with the greatest production globally,and in the developing world by 2025 (Rosegrant et al., 2008) In large parts

of Africa, maize is the principle staple crop; accounting for an average of32% of consumed calories in Eastern and Southern Africa, rising to 51% insome countries (Table 1).Heisey and Edmeades (1999)estimated that one-quarter of the global maize area is affected by drought in any given year.Additional constraints causing significant yield and economic losses annuallyinclude low soil fertility, pests, and disease It is difficult to give an accuratefigure on combined maize yield losses due to these stresses; however, it islikely to be extensive Maize yields remain low and highly variable betweenyears across sub-Saharan Africa at 1.6 t ha
1, only just enough to reach self-sufficiency in many areas (Ba¨nziger and Diallo, 2001; FAOSTAT, 2010).The world population is expected to surpass 9 billion by 2050, withpopulation growth highest within developing countries Harvest at currentlevels of productivity and population growth will fall far short of futuredemands Projections of climate change will further exacerbate the ability toensure food security within many maize producing areas The development

of improved germplasm to meet the needs of future generations in light ofclimate change and population growth is of the upmost importance(Easterling et al., 2007)

Population (thousands)a

Total areab(ha)

% of total calorie intake from maize consumptionb

Population (thousands)a

Total areab(ha)

% of total calorie intake from maize consumptionb

Past experience has demonstrated that the use of new varieties side improved management options can offset yield losses by up to 40%(Thornton et al., 2009) The development and application of moleculartools in plant breeding started in the early 1980s Molecular breedingoffers the ability to increase the speed and efficiency of plant breeding(Whitford et al., 2010) In rice, SUB1 a major QTL (quantitative traitloci) controlling submergence tolerance was recently identified andintrogressed into local mega varieties using only two backcrosses andone selfing generation (Septiningsih et al., 2009) In maize, a geneencoding b-carotene (crtRB1) was recently identified and is nowbeing introgressed into tropical germplasm using marker-assisted selec-tion (MAS) to alleviate vitamin A deficiency in the developing world(Yan et al., 2010) Many more examples of the use of molecular tools toquickly develop improved germplasm with resilience to major abioticand biotic stress are beginning to emerge As the impacts of climatechange will vary regionally and given the time lag between the devel-opment of improved germplasm and adoption in farmers’ fields, there is

along-an immediate need to identify future breeding target environments along-andreduce uncertainty within climate projections to allow priority settingfor both researchers and policy markers

This review addresses the potential impacts of climate change on maizeproduction with specific reference to sub-Saharan Africa Considerable gapsremain in our knowledge of how agricultural systems will be affected.Earlier climate projections have tended to focus at the country level.While these studies have helped to increase our understanding of potentialfuture climates, at such low resolution priority setting of agriculturalresearch is not possible Climate projections for sub-Saharan Africa at themaize mega-environment level within countries are presented Currentresearch and potential new tools to increase maize resilience to abiotic andbiotic stresses are presented Finally, mitigation technologies and practicesfor maize-based systems are discussed

2 Likely Climate Scenarios for

Sub-Saharan Africa and South Asia

and Identification of Hot Spots

Previously climate projections were developed using the outputs offew global climate models (GCMs) at low resolution Large variation existswithin the outputs of GCMs and for regional application the use of multiplemodels reduces the error in both the mean and variability Additionally, theearlier focus on low resolution modeling at the country level masks large

variation in key factors, such as climate and topography, and reduces thepotential application of projections as decision making tools for identifyingpriority areas for research Working at the regional level, Thornton et al.(2009)showed large spatial variation in simulated yield production changes

of maize and beans within the highlands of Ethiopia and Kenya There is apressing need to identify future breeding targets and hot spots of vulnera-bility to climate change in maize growing areas

The CIMMYT maize breeding program is organized around the cept of mega-environments, or areas with broadly similar environmentalcharacteristics with respect to maize production, to target its breedingprograms Mega-environments were delineated using environmental factors(maximum temperature, rainfall, and sub-soil pH) as explanatory factors forgenotype by environment interaction of advanced hybrids from multi-environmental trials (Ba¨nziger et al., 2006; Setimela et al., 2005) Similarcombinations of climatic and edaphic conditions exist within and acrosscontinents, allowing maize mega-environments to be approximately iden-tified on the basis of GIS data Six maize mega-environments were identi-fied across sub-Saharan Africa (Fig 1) and South and South-East Asia(Fig 2), respectively Germplasm developed at key sites within mega-environments should have broad adaptation across the mega-environment

con-As climatic conditions change at particular experimental sites and maizeproducing regions, mega-environment assignments will need to be re-assessed to guide breeders to appropriate new germplasm and target envir-onments CIMMYT’s global maize breeding programs can rapidly sourceelite, potentially useful germplasm from the full range of mega-environ-ments in the developing world Although it should be noted that end-usecharacteristics, color preferences, and other factors may often prevent thedirect substitution of, say, lowland-adapted varieties for varieties in mid-elevation mega-environments that are experiencing warming Thus, inaddition to being able to source germplasm from mega-environmentswith conditions similar to those arising from climate change in their ownareas, breeders will need the capacity to rapidly move stress tolerance traitsinto germplasm preferred by people in the target environment they serve.Previous research strongly suggests maize growing regions of sub-Saharan Africa will encounter increased growing season temperatures andfrequency of droughts (IPCC, 2007) To establish changes in maximumtemperatures and annual rainfall difference at the maize mega-environmentlevel within countries, downscaled outputs from 19 SRES (Special Report

on Emissions Scenarios) models and the A2 emissions scenario with dataprovided by CIAT (Ramirez and Jarvis, 2008) were used with the followingclimate change models: BCCR-BCM 2.0, CCCMA-CGM2, CCCMA-

IAP-FGOALS-1.0G, GISS-AOM, GFDL-CM2.1, GFDL-CM2.0, MK3.0, IPSL-CM4, MIROC 3.2-HIRES, MIROC 3.2-MEDRES,

CSIRO-MIUB-ECHO-G, MPI-ECHAM5, CSIRO-MIUB-ECHO-G, MPI-ECHAM5,MRI-CGCM2.3.2A., NCAR-PCM1, NIES99, UKMO-HADCM3.Countries were subdivided into maize mega-environments as shown inFig 1 For temperature and precipitation projections the period 2040–

2069 was selected, average temperatures and annual precipitation duringthis period are presented and referred to as 2050 Climatic data was down-scaled to approximately 5 m resolution and the relationship between histor-ical climate data from meteorological stations and climate model outputswas established using an empirical statistical approach Average temperatureswere derived from the combined outputs of all 19 models using ArcGISsoftware (Ormsby et al., 2009) The differences between future predictionsand current long-term average values (1950–2000) were calculated usingthe worldClim 1.4 dataset also at 2.5 min resolution as a reference (Hijmans

et al., 2005) Values within mega-environments within the respectivecountries were averaged

Wet lower mid-altitude

Wet upper mid-altitude

Figure 1 Maize mega-environments within sub-Saharan Africa (adapted from Hodson

et al., 2002a ).

The results of temperature simulations for 2050 across maize environments within sub-Saharan Africa show a general trend of warming,

mega-in agreement with previous projections conducted at the country level(Burke et al., 2009; IPCC, 2007) (Fig 3) In sub-Saharan Africa, warming

is the greatest over central southern Africa and western semi-arid margins ofthe Sahara and least in the coastal regions of West Africa Maximumtemperatures are predicted to increase by 2.6 C, with the increase inminimum temperatures slightly lower, with an average of 2.1 C

In agreement with Burke et al (2009), the range of temperatures within acountry is likely to be larger than the range of temperatures across years(2010–2050) Average optimum temperatures in temperate, highland trop-ical, and lowland tropical maize lie between 20–30 C, 17–20 C, and30–34 C, respectively (Badu-Apraku et al., 1983; Brown, 1977; Chang,1981; Chowdhury and Wardlaw, 1978) Maximum temperatures currentlyexceed optimal temperature conditions for lowland tropical maize (34C)within several countries (Burkina Faso, Chad, Eritrea, Gambia, Mali,Mauritania, Niger, Nigeria, Senegal, and Sudan), although the area ofmaize grown within several of these regions is small Maize is an important

Wet lower mid-altitude

Wet upper mid-altitude

Figure 2 Maize mega-environments within Asia (adapted from Hodson et al., 2002b ).

crop in the highlands of Kenya, Ethiopia, and Tanzania Average tures within these regions are currently at the threshold for highland maizeand will likely exceed this threshold by 2050.

tempera-Projections of changes in precipitation show a general trend of increasedannual precipitation in western and eastern Africa In general, annualprecipitation is projected to decrease within Malawi, Madagascar, north-east South Africa, Angola, Gabon, Cameroon, and Congo Annual rainfall

in Cameroon, Congo, and Gabon is relatively high with an average of 1504,

1475, and 1564 mm rainfall annually, respectively (calculated from 1995 to

2005 rainfall data fromMitchell and Jones, 2005) Therefore, the decrease inrainfall may not have a major impact on maize production within thesecountries Decreasing precipitation combined with increasing temperaturesmay have major implications for maize production within Mozambique,South Africa, and Madagascar These results highlight potential hotspots fortargeting research; however, further refinement is required to decipherpotential changes in precipitation during the growing season (particularlyduring the reproductive stage) and potential impacts of combined changes

Difference max temp

sub-including heat and drought stress combined Given the projected changes intemperature and precipitation, two of the main environmental factors used

to delineate current maize mega-environments, it is likely some regions willhave to be reclassified into new mega-environments or a new environmen-tal classification system developed Ortiz et al (2008)previously examinedpotential changes in major wheat production environments as a result ofclimate change using one GCM The results of their study suggest up to51% of the wheat regions within the Indo-Gangetic Plains would need to bereclassified (Fig 4)

Annual Rainfall differences

3 Adaptation Technologies and Practices

for Addressing Near-Term and

Progressive Climate Change

3.1 Abiotic stresses—Drought, heat, and waterlogging

3.1.1 Drought

Drought is a widespread phenomenon across large areas of sub-SaharanAfrica, with an estimated 22% of mid-altitude/subtropical and 25% oflowland tropical maize growing regions affected annually inadequatewater supply during the growing season (Heisey and Edmeades, 1999) InEastern and Southern Africa, a general relationship can be observedbetween annual rainfall and national average maize yields (Fig 5)(Ba¨nziger and Diallo, 2001) Conventional drought stress tolerance breed-ing has yielded significant dividends in maize (Ba¨nziger et al., 2006).Conventional breeding for drought tolerance has resulted in gains of up

to 144 kg ha
1yr
1in tropical maize when stress was imposed at flowering(Edmeades et al., 1999) In temperate maize, the rate of breeding progresshas been estimated at 73 kg ha
1 yr
1 for mild stress (Duvick, 1997),

146 kg ha
1yr
1when the stress was imposed at the flowering stage, and

400

1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005

Rainfall Yield

Figure 5 Relationship between rainfall and average maize yields across Eastern and Southern Africa (adapted from B€anziger and Diallo, 2001 ) Data source: FAOSTAT (2010) and Mitchell and Jones (2005)

76 kg ha
1yr
1when the stress was imposed during mid-grain filling stage(Campos et al., 2004) Success in breeding drought-tolerant tropical maize,has been largely attributed with the application of proven drought breedingmethodologies in managed stress screening (Ba¨nziger et al., 2006).

While drought negatively affects all stages of maize growth and duction the reproductive stage, particularly between tassel emergence andearly grain filling, is the most sensitive to drought stress (Grant et al.,

pro-1989) Drought stress during this period results in a significant reduction

in grain yield, associated with a reduction in kernel size (Bolan˜os andEdmeades, 1993a,b) The susceptibility of maize to drought stress isgenerally attributed to the separation of its male and female flowers(Grant et al., 1989) While silking is delayed under drought stress, there

is little effect on the timing of pollen shed Comparisons of the responses

of male and female reproductive tissues under drought stress confirmedfemale tissues to be the most sensitive (Herrero and Johnson, 1980; Mossand Downey, 1971) Westgate and Boyer (1986) compared the response

of male and female reproductive tissues and found silk water potential tofollow changes in leaf water potential, while pollen water potentialremained unchanged The results of their experiments indicated stigmatictissues were in moderate hydraulic contact with vegetative tissue Usingstem infusions of sucrose solution,Boyle (1990)showed that the effects ofdrought at flowering could be partially alleviated; suggesting silk delaymay be a symptom of limited assimilates supply rather than a primarycause of bareness The delay in silking results in decreased male–femaleflowering synchrony or increased anthesis-silking interval (ASI) Earlyfield experiments reported an 82% reduction in grain yield as ASIincreased from 0 to 28 days (DuPlessis and Dijkhuis, 1967, as reported

in Edmeades et al., 1993)

In the 1970s, CIMMYT initiated a drought breeding program for maizeusing the elite lowland tropical maize population “Tuxpen˜o Sequia”(Bolan˜os and Edmeades, 1993a,b; Bolan˜os et al., 1993) A recurrent selec-tion approach was applied to increase the frequency of alleles conferringtolerance Evaluations were conducted under managed drought stressimposed at flowering with selection for grain yield, increased floweringsynchrony, and delayed leaf senescence (Bolan˜os and Edmeades, 1993a).Drought stress reduced grain yield by an average of 15–30% relative to thewell-watered control Over eight cycles of full-sib recurrent selection thedrought tolerance of Tuxpen˜o Sequia was improved Selection gains wereassociated with reduced ASI, fewer barren plants, a smaller tassel size, agreater harvest index, and delayed leaf senescence, with no changes in wateruptake or biomass observed (Bolan˜os and Edmeades, 1993a,b; Bolan˜os et al.,1993; Chapman and Edmeades, 1999) Root biomass decreased by one-third in the top 50 cm (Bolan˜os et al., 1993) Retrospective studies intemperate maize hybrids selected to represent yield improvements from

1950s to 1980s (Tollenaar and Lee, 2006; Tollenaar and Wu, 1999) showedyield Yield were associated with more efficient resource capture and use ofresources, particularly under stress.

New secondary traits and phenotyping methods will help the success ofdrought tolerance breeding for tropical maize to continue Yield is afunction of many processes throughout the plant cycle thus integrative traitsthat encompass crop performance over time or organization level (i.e.,canopy level) will provide a better alternative to instantaneous measure-ments which only provide a snapshot of a given plant process (Araus et al.,

2008) Many new phenotyping tools based on remote sensing are nowavailable including nondestructive measurements of growth-related para-meters based on spectral reflectance (Marti et al., 2007) and infrared ther-mometry to estimate plant water status (Jones et al., 2009) Recently,Cabrera-Bosquet et al (2009a,b) proposed oxygen isotope enrichment(D18

O) and kernel ash content as new physiological traits to improvemaize yields in drought-prone environments Both traits provide an inte-grative measurement of physiological traits during the crop growth cycle,with D18

O reflecting plant evaporative conditions throughout the cropcycle (Barbour et al., 2000) while kernel ash content provides information

on integrative photosynthetic and retranslocation processes during grainfilling (Araus et al., 2001) Together these tools have potential to be used

in the characterization and identification of key drought tolerant donors to

be used in breeding programs However, further work is required toevaluate their possible application as selection tools within drought breedingprograms

3.1.2 Heat

By the end of this century, growing season temperatures will exceed themost extreme seasonal temperatures recorded in the past century (Battistiand Naylor, 2009) Using crop production and meteorological records,Thomson (1966) showed that a 6 C increase in temperature during thegrain filling period resulted in a 10% yield loss in the U.S Corn Belt A laterstudy in the same region showed maize yields to be negatively correlatedwith accumulated degrees of daily maximum temperatures above 32 Cduring the grain filling period (Dale, 1983) Lobell and Burke (2010)suggested that an increase in temperature of 2C would result in a greaterreduction in maize yields within sub-Saharan Africa than a decrease inprecipitation by 20% A recent analysis of more than 20,000 historicalmaize trial yields in Africa over an 8-year period combined with weatherdata showed for every degree day above 30C grain yield was reduced by1–1.7% under optimal rainfed and drought conditions, respectively (Lobell

et al., 2011) These reports highlight the need to incorporate tolerance toheat stress into maize germplasm However, relatively little research hasbeen conducted on heat stress compared to other abiotic stresses in maize

(Paulsen, 1994) The vast majority of heat stress research has been ducted on temperate maize germplasm for high production areas There-fore, limited breeding progress has been made in the development ofimproved maize germplasm with specific tolerance to elevated tempera-tures Heat stress can be defined as temperatures above a threshold level thatresults in irreversible damage to crop growth and development and is afunction of intensity, duration, and the rate of increase in temperature.Further, different plant tissues and organs, and different developmentalstages are affected by heat stress in different ways, depending on thesusceptibility of the dominant metabolic processes that are active at thetime of stress (Larkindale et al., 2005) Accumulated or acute high tempera-tures can cause an array of morphological, anatomical, physiological, andbiochemical changes within maize The threshold temperature for maizevaries across environments as previously described in Section 2 The mostsignificant factors associated with maize yield reduction include shortenedlife cycle, reduced light interception, and increased sterility (Stone, 2001).

con-To stabilize maize yields under elevated temperatures it is necessary tounderstand the mechanisms responsible for yield loss

The temperature threshold for damage by heat stress is significantlylower in reproductive organs than in other organs (Stone, 2001) Successfulgrain set in maize requires the production of viable pollen, interception ofthe pollen by receptive silks, transmission of the male gamete to the egg cell,initiation and maintenance of the embryo and endosperm development(Schoper et al., 1987a,b) High temperature during the reproductive phase

is associated with a decrease in yield due to a decrease in the number ofgrains and kernel weight Under high temperatures, the number of ovulesthat are fertilized and develop into grain decreases (Schoper et al., 1987a,b)

A comparison of the response of male and female reproductive tissues toheat stress demonstrated that female tissues have greater tolerance (Dupuisand Dumas, 1990) Pollen production and/or viability have been high-lighted as major factors responsible for reduced fertilization under hightemperatures Pollen produced under high temperature has reduced viabil-ity and in vitro germination (Dupuis and Dumas, 1990; Herrero andJohnson, 1980; Schoper et al., 1986, 1987a,b) Additionally, high tempera-tures are responsible for reduced pollen water potential, quantity of thepollen shed, and pollen tube germination (Dupuis and Dumas, 1990;Schoper et al., 1987a,b) Pollen desiccated to 20% of its original watercontent is still capable of germination (Barnabas et al., 2008); thus, thereduction in pollen water potential under heat stress is unlikely to bethe cause of reduced pollen viability (Schoper et al., 1987b) The location

of the tassel also provides maximum exposure to extreme temperatures,increasing the probability of pollen damage as a result of heat stress.High temperature during the early stages of kernel development has adetrimental effect on kernel development and final kernel mass due to a

reduction in the number and/or size of endosperm cells formed therebyreducing sink capacity (Jones et al., 1984) During this stage heat stressaffects cell division, sugar metabolism, and starch biosynthesis, reducingsubsequent dry matter accumulation within kernels (Commuri and Jones,2001; Engelen-Eigles et al., 2000; Monjardino et al., 2005) The duration ofthe grain filling process (ca 35 days) is the longest physiological processduring the reproductive stage, increasing the probability of experiencinghigh temperature during this stage Maize kernel weight is the product ofthe rate and duration of grain filling, both of which are affected by temper-ature High temperature during this period is associated with a reduction inthe duration of grain filling (Badu-Apraku et al., 1983; Hunter et al., 1977;Muchow, 1990) Earlier studies showed temperature to increase the growthrate of kernel development (Muchow, 1990; Singletary et al., 1994); how-ever, this increase was unable to compensate for the reduction in growthduration and this resulted in kernels that weigh less (Singletary et al., 1994).When the rate and duration of grain filling are calculated on the basis

of accumulated heat units, the greatest reduction is in the rate, andnot the duration of grain filling Thus, the larger reduction in the rate ofgrain filling was responsible for the heat-related reduction in seed mass(Wilhelm et al., 1999)

Grain filling duration is determined by a number of factors includingsucrose availability and the activity of starch and sugar metabolism enzymes

in the kernel (Jones et al., 1984) Heat stress during grain filling reducesendosperm starch content, the primary constituent of kernels (Singletary

et al., 1994).Cheikh and Jones (1994)studied the effect of heat stress (35C)

on sink activity of maize kernels in vitro Heat stress was not associated withreduced carbon supply to the kernel, suggesting that the effect of heat stresswas related to changes in carbon utilization and partitioning Thus, heatstress did not reduce sink activity by reducing kernel uptake of sugars but byadversely affecting the conversion of sugars to storage products In vitrostudies on the effects of high temperature on carbohydrate metabolismenzymes in maize kernels suggest ADP glucose pyrophosphorylase andsucrose synthase to be the most sensitive with developmental peaks ofactivity similar to profiles of starch accumulation (Keeling et al., 1994;Singletary et al., 1994; Wilhelm et al., 1999)

Elevated temperatures also negatively affect the seedling and vegetativestages During the autotrophic phase of germination, plant energy is directlyaffected by soil temperature (Stone, 2001) High temperature reduces bothseedling percentage and growth (Weaich et al., 1996a) In maize, seedlinggrowth is maximized at a soil temperature of 26C and above this temper-ature, root, and shoot mass both decline by 10% for each degree increaseuntil 35 C when growth is severely retarded (Walker, 1969) Reducedseedling growth has been suggested to be associated with poor reservemobilization, with reduced protein synthesis observed in seedlings grown

under elevated temperatures (Riley, 1981) Seedlings growing in high soiltemperatures are likely to suffer further damage as the associated slowergrowth rate delays canopy closure, consequently reducing soil shading.Above 35 C, maize leaf elongation rate, leaf area, shoot biomass, andphotosynthetic CO2 assimilation rate decrease (Watt, 1972) Elongation

of the first internode and overall shoot growth of maize has been suggested

as the most sensitive processes of the vegetative stage to high temperatures(Weaich et al., 1996b) C4 plants have a higher optimum temperature forphotosynthesis compared to C3 plants due to the operation of a CO2-concentrating system that inhibits rubisco oxygenase activity (Berry andBjo¨rkman, 1980) However, a comparison of the photosynthetic responsesand sensitivity of the light reactions in both C3and C4crop plants subjected

to brief heat stress suggested that the C4pathway alone did not necessarilyconfer tolerance to high temperature (Ghosh et al., 1989) Differences inphotosynthetic response were more closely associated with light reactions,particularly the sensitivity of photosystem II activity under elevatedtemperatures

Research to date on specific tolerance to heat stress in maize has mainlyfocused on biochemical and molecular responses using only a limited number

of accessions and heat stress applied in vitro as a single, rapid heat stress event

In wheat, progressive heat stress has a more deleterious effect on yield andyield components when compared to a single, rapid event of heat stress(Corbellini et al., 1997) In maize, no comparisons have been made betweenrapid heat treatments (in vitro and field) and progressive heat stress, as com-monly experienced in the field Given that different traits and mechanisms arelikely to provide adaptation for different types of heat stress (i.e., varying induration, intensity, and timing); heat stress environments need to be defined

to enable the assessment of the relevance of individual physiological andbreeding experiments for the target populations of environments

3.1.3 Waterlogging

Over 18% of the total maize production area in South and Southeast Asia isfrequently affected by floods and waterlogging problems, causing produc-tion losses of 25–30% annually (Zaidi et al., 2010) (Fig 6) Although the area

of land in sub-Saharan Africa affected by waterlogging is lower than in Asia,

it is a risk in a few areas (Fig 7) Waterlogging stress can be defined as thestress inhibiting plant growth and development when the water table of thesoil is above field capacity The diffusion rate of gases in the flooded soilcould be 100 times lower than that in the air, leading to reduced gasexchange between root tissues and the atmosphere (Armstrong and Drew,

2002) As a result of the gradual decline in oxygen concentration within therhizosphere, the plant roots suffer hypoxia (low oxygen), and duringextended waterlogging (more than 3 days) anoxia (no oxygen) (Zaidi

et al., 2010) Carbon dioxide, ethylene, and toxic gases (hydrogen sulfide,

ammonium, and methane) also accumulate within the rhizosphere duringperiods of waterlogging (Ponnamperuma, 1984) A secondary effect ofwaterlogging is a deficit of essential macronutrients (nitrogen, phosphorous,and potassium) and an accumulation of toxic nutrients (iron and magne-sium) resulting from decreased plant root uptake and changes in redoxpotential Nutrient uptake is reduced as a result of several factors Anaerobicconditions reduce ATP production per glucose molecules, thereby reduc-ing energy available for nutrient uptake Reduced transport of water furtherreduces internal nutrient transport Reduced soil conditions decrease theavailability of key macro nutrients within the soil Under waterloggingconditions nitrate is reduced to ammonium and sulfate is converted tohydrogen sulfide, and both become unavailable to most of the non-wetlandcrops, including maize Availability of phosphorous may increase ordecrease depending upon soil pH during waterlogging.

The extent of damage due to waterlogging stress varies significantly withthe developmental stage of the crop Previous studies have shown that maize

is comparatively more susceptible to waterlogging from the early seedlingstage to the tasseling stage (Mukhtar et al., 1990; Zaidi et al., 2004) Theeffects of waterlogging result in a wide spectrum of changes at the molecu-lar, biochemical, physiological, anatomical, and morphological levels, and

Percent of area with waterlogging problems 1–20

21–40 41–60 61–80 81–100 Figure 6 Waterlogging risk in Asia Data source: Hodson et al (2002a), Sanchez et al (2003), You et al (2000, 2006)

such changes have been extensively reviewed (Kennedy et al., 1992; Perataand Alpi, 1993; Ricard et al., 1994) The first symptoms of waterlogging areleaf rolling and wilting and reduced stomatal conductance These changesare followed by root growth inhibition, changes in root, and shoot mor-phology, change in root to shoot ratio, leaf senescence, and brace rootdevelopment by above ground nodes (Rathore et al., 1998; Zaidi and Singh,

2001, 2002; Zaidi et al., 2003) Rapid wilting is related to water deficit due

to net loss of water from shoot, which might be related to increasedresistance to water flow in roots (Levitt, 1980) In maize, decrease inwater availability under waterlogging was found to be associated with rootdecay and wilting Reduced stomatal conductance and high humiditycauses a reduced demand on the root system for water acquisition Leach-ing-induced disturbance in the osmotic gradient of the root cortex results ininhibition of radial movement of water from root hairs across the cortexinto xylem Consequently, the water supply to above ground plant parts isreduced and plants suffer internal drought stress

Percent of area with

Figure 7 Waterlogging risk in Africa Data source: Hodson et al (2002a), Sanchez

et al (2003), You et al (2000, 2006)

A sharp decline in aerobic respiration in root tissues is one of the earliestresponses of plants under waterlogging Waterlogging-induced anaerobiosisresults in energy starvation, with only 2 ATP produced per mole of glucose,coupled with the production of toxic end products (ethanol, lactate, malate,alanine).Zaidi et al (2003)found that NADþ-alcohol dehydrogenase activityincreased exponentially in the tolerant maize genotypes under waterloggingwith a decline in ADH-activity in sensitive genotypes.Sachs (1993)analyzedwaterlogging tolerance in maize and found that ADH-activity was apparentwithin 90 min and reached its highest level after approximately 5.0 h of theanoxia treatment They concluded that variation in the stress tolerance wasrelated the ADH-activity However,Liu et al (1991)suggested that increasedalcoholic fermentation was a temporary adaptation and a major cause of rootinjury during flooding, and flooding tolerance was related to low ethanolfermentation Liao and Lin (1995) also suggested that ADH activity waspositively correlated with the magnitude of excess moisture injury, and geno-types with higher ethanol production were less tolerant to flooding It has beenproposed that ethanol accumulation may have a “self poisoning” role in flood-intolerant plants Plant roots under waterlogging conditions require a largeamount of carbohydrate due to inefficient anaerobic respiration Increasedanerobic respiration results in rapid depletion of carbohydrate in roots, causing

“carbohydrate starvation” during periods of waterlogging (Setter et al., 1987).Poorly developed brace roots before tasseling have been suggested as animportant factor for increased susceptibility during the vegetative growth(Rathore et al., 1998; Zaidi et al., 2003) At later growth stages, some genotypeshave the ability to produce adventitious roots with aerenchyma formation inthe cortical region, thereby increasing the ability to tolerate excess water withinthe rhizosphere (Rathore et al., 1998; Zaidi, 2003) Under extended water-logging (>3 days) formation of lysigenous aerenchyma in the cortical region ofroots and brace root development on above ground nodes has been observed inwaterlogging tolerant maize genotypes (Mano and Omori, 2007; Mano et al.,2005; Rathore et al., 1998; Zaidi and Singh, 2001, 2002; Zaidi et al., 2003) Inmaize, production of adventitious roots with aerenchyma is not a constitutivebut an adaptive trait, particularly under waterlogging conditions Aerenchymaare formed through ethylene-induced cell lysis, a process of progressive celldeterioration or precocious senescence (Jackson, 1989, 1990; Vartapetian andJackson, 1997) Aerenchyma provide a diffusion path of low resistance for thetransport of oxygen from aerial parts of the newly developed brace root to theroots present under severe anoxic conditions (Kawase and Whitmoyer, 1980;Laan et al., 1989) They also provide a path for diffusion of volatile compoundssuch as ethylene, methane, CO2, ethanol, and acetaldehyde (Vartapetian andJackson, 1997; Visser et al., 1997)

Significant genotypic variation has been observed for tolerance to ing in maize (Rathore et al., 1998; Zaidi and Singh, 2001; Zaidi et al., 2003).This variability could be exploited to develop maize varieties tolerant to

flood-intermittent waterlogging stress during the summer-rainy season in thetropics In the 1980s, EMPBRAPA in Brazil initiated a breeding programfor waterlogging tolerance in maize (Ferreira et al., 2007) Recurrentselection over 12 cycles resulted in the development and subsequent release

of the waterlogging tolerant BRS 4154 maize line, with a 20% yieldadvantage under waterlogging compared to the original source The results

of this long-term breeding effort highlight the potential to developimproved maize germplasm with tolerance to waterlogging and, in addi-tion, the time investment required under conventional breeding

3.2 Biotic stresses of maize under the changing climate

Abiotic stresses account for a significant proportion of maize yield lossesworldwide The predominant insect-pests and diseases vary across environ-ments (Table 2) and a major challenge in adapting crops to climate changewill be the maintenance of genetic resistance to pests and diseases (Reynoldsand Ortiz, 2010) Changing climates will affect the diversity and respon-siveness of agricultural pests and diseases Studying and understanding thedrivers of change will be essential to minimize the impact of plant diseasesand pests on maize production

Table 2 Major biotic stresses associated with maize production losses in Asia, Africa, and Latin America

Downey mildew Borer (Chilo, Sesamia spp.)

Borers (Chilo, Sesamia spp.)

Downy mildew Borers (Chilo, Sesamia spp.) Sub-Saharan

Africa

Turcicum blight Common rust Ear rots

Gray leaf spot Streak virus Ear rots Weevils Borers (Chilo, Sesamia spp.)

Striga Streak virus Borers

Latin America

and Caribbean

Ear rots Rust Turcicum blight

Turcicum blight Borer (S W corn borer)

Tar spot complex Ear rots

Gray leaf spot

Fall armyworm Corn stunt complex Ear rots

Gray leaf spot

3.2.1 Plant diseases

For a disease to occur a virulent pathogen, susceptible host, and favorableenvironment are essential (Legre`ve and Duveiller, 2010) All of these com-ponents are strongly coupled with environmental conditions Global climatechanges have the potential to modify host physiology and resistance, and alterboth stages and rates of pathogen development Environmental conditionscontrolling disease development include rainfall, relative humidity, tempera-ture, and sunlight Changes in these factors under climate change are highlylikely to have an effect on the prevalence of diseases and emergence of newdiseases For example, in Latin America tar spot complex, caused by Phylla-chora maydis Maubl., Monographella maydis Mu¨ller & Samuels and Coniothyriumphyllachorae, was previously rare However, recent epidemics of the tar spotcomplex have been recorded in Guatemala, Mexico, Colombia, and ElSalvador due to recent climate variability (Pereyda-Herna´ndez et al., 2009).The disease infection cycle includes inoculum survival, infection,latency period, production of new propagules, and dispersal, all of whichare strongly influenced by environmental conditions The penetration orinfection of a plant by infectious propagules is determined by specificenvironmental conditions In general, fungi require high relative humidity

or moist leaf surfaces for infection; changes in these conditions will increaseinfection rates For example, Cercospora zeae-maydis and Cercospora zeinacause gray leaf spot (GLS) in maize and are highly sensitive to environmentalconditions (Crous et al., 2006) Under dry conditions (relative humidity

<80%), the pathogen ceases to grow and infection stops (Thorson andMartinson, 1993) Therefore, changes in temperature, humidity, and rain-fall patterns have the potential to increase infection by many maize patho-gens Increased temperature reduces the latency period (generation time)resulting in a higher number of generations per season Generation timedetermines the amplification of plant disease in two ways—accelerating andincreasing inoculums load and/or affecting pathogen evolution rates and apathogen’s capacity to adapt to the environment—potentially allowing thepathogen to adapt faster to the environment than the host

Climate change may also affect gene flow, the process through whichparticular alleles or individuals are exchanged among separate populations.This will increase pathogen population diversity leading to variation in hostresistance, variation in pathogen virulence, and new specific interactions Thishas the potential to result in new diseases or pathogen emergence, and theintroduction of pathogens into new ecological niches Depending on thedistribution of populations and environmental conditions that are influenced

by climate change, gene flow leads to an increase in population diversity or tothe introduction of a new population in new ecological niches

An important example of changes in growing season conditions beinglinked to outbreaks of diseases, with serious human health implications, ismycotoxins and their prevalence within maize systems Mycotoxins are toxic

secondary fungal metabolites that contaminate agricultural products andthreaten food safety Different groups of mycotoxins are produced by differ-ent fungi A flavus and A parasiticus produce aflatoxin, F verticillioides pro-duces fumonisin, and F graminierum produces deoxynivelanol (DON) andzearalenone) (Cardwell et al., 2001; Miller, 2008) Mycotoxin contamination

is a serious problem with long-term consequences for human and animalhealth Sublethal exposure to mycotoxins suppress the immune system,increase the incidence and severity of infectious diseases, reduce child growthand development, and reduce the efficacy of vaccination programs (Williams

et al., 2004) Consumption of high doses of mycotoxins causes acute illnessand can prove fatal In 2004, more than 125 people died in Kenya from eatingmaize with aflatoxin B1 concentrations as high as 4400 parts per billion—220times the Kenyan limit for foods (Lewis et al., 2005) The maize implicated inthis outbreak was harvested during unseasonable early rains and stored underwet conditions conducive to mold growth and therefore aflatoxin contami-nation (CDC, 2004) Previous outbreaks in Kenya and India have also beenattributable to unseasonable, heavy rain during harvest (Krishnamachari et al.,1975; Ngindu et al., 1982) Environmental conditions conducive to myco-toxin producing fungi vary A flavus competes poorly under cool conditionsand the prevalence of A flavus is higher in warmer environments (above

25C) compared to cooler environments (20–25C) (Shearer et al., 1992).The environment influences not only the quantity of aflatoxin producers butalso the “type” of producer present (Horn and Dorner, 1999) In Africa, the

“S” morphotypes of A flavus are associated with hot and dry “agro-ecologicalzones” with latitudinal shifts in climate influencing fungal community struc-ture (Cardwell and Cotty, 2002) For the Fusariums, F graminearum ispredominate in temperate maize growing environments, whereas F verticil-lioides and F proliferatum and fumonisins are more widely spread in tropicaland subtropical environments (Miller, 1994) The optimal temperature rangefor F graminearum is between 24 and 28C and above this temperature range

F verticillioides out-competes F graminearum (Miller, 2001; Reid et al., 1999).Increasing temperatures within maize growing regions are highly likely tochange the geographical distribution and predominance of F verticillioides,particularly in currently cooler regions where it will replace F graminerum.This shift in Fusarium species will result in a change in mycotoxins, fromdeoxynivalenol and zearalenone (produced by F graminierum) to fumonisin(produced by F verticillioides) Increased incidence of F verticillioides andsubsequent fumonisin contamination has already been reported in Guatemala,Mexico, Zimbabwe, and Kenya (Torres et al., 2007)

3.2.2 Insect-pests

The dynamics of insect-pests are also strongly coupled with environmentalconditions Insects do not use their metabolism to maintain their bodytemperature, and are dependent on ambient temperature to control their

body temperature Temperature is therefore the single most importantenvironmental factor influencing insect behavior, distribution, develop-ment and survival, and reproduction Insect life stage predictions are calcu-lated on accumulated degree days, which is a function of both time andtemperature Increased temperature can speed up the life cycle of insectsleading to a faster increase in pest populations It has been estimated that a

2 C increase in temperature has the potential to increase the number ofinsect life cycles during the crop season by one to five times (Bale et al.,2002; Petzoldt and Seaman, 2005; Porter et al., 1991) The feeding rate ofmany arthropod vectors increases at higher temperatures, thus increasingexposure of crops to mycotoxigenic fungi thereby increasing the spread ofmycotoxins (Bale et al., 2002; Dowd, 1992)

Insect damage has been shown to be closely related to Fusarium orAspergillus ear rots (Miller, 2001; Munkvold and Hellmich, 2000) A fieldsurvey in Austria demonstrated that the incidence of the European maizeborer increased F verticillioides disease and fumonisin concentrations but not

F graminierum (Lew et al., 1991) Therefore, the increased global warmingand drought incidences will favor insect proliferation and herbivory, whichwill likely increase the incidence and severity of insect related damages aswell as aflatoxin and fumonisin mycotoxins in maize Higher averagetemperatures have the potential to change the geographical distribution ofcrops This may in turn result in an expansion of the geographical distribu-tion of insect-pests and their associated pathogens (e.g., maize streak virus(MSV), corn stunt complex that are vectored by different species of leafhoppers), resulting in a change in the geographical distribution of diseases

3.3 Strategies for mitigating climate-related effects of

biotic stresses on maize yields

Breeding for disease and insect resistance requires an understanding ofparasite biology and ecology, disease cycles and drivers influencing theevolution of plant–pathogen interactions, because unlike abiotic stresses,biotic stress resistance is influenced by genetic variability in the pest/patho-gen population As a result of the evolving pest/pathogen populations andthe changes in fitness favoring new pathotypes/biotypes, improving resis-tance to biotic stresses has been a long-term focus of agricultural researchers.The long-term success of breeding for disease or insect-pest resistance willdepend on a more in-depth and clear understanding of (i) the nature of thepathogen/insect-pest, and diversity of virulence in the populations; (ii) theavailability, diversity, and type of genetic resistance; (iii) availability ofsuitable sites (hot spots), screening methodologies/protocols for generatingadequate disease/insect-pest pressures, and tracking resistance; (iv) selectionenvironments and methodologies for rapidly generating multiple stressresistant inbred lines, and their use in hybrid or variety development

Significant progress has been made over the decades in the identification

of stable genetic resistance for major maize diseases (Bosque-Perez, 2000;Dowswell et al., 1996; McDonald and Nicol, 2005; Pratt and Gordon,2006; Welz and Geiger, 2000) However, the population structure ofmost maize pathogens remains inadequately characterized Also, concertedefforts are required to widely test the available sources of resistance inmultiple and relevant environments to expose them to a wide spectrum ofpathogen strains and to facilitate identification of the most suitable resistancegenes/alleles for use in the breeding programs Research at CIMMYT isfocused on multi-location phenotyping of a common set of 500 maizeinbred lines for some prioritized diseases, namely GLS, TLB (turcicumleaf blight), MSV, and ear rots, across more than 15 locations in sub-SaharanAfrica, Latin America, and Asia This will help identify stable sources ofresistance to key diseases and identify key phenotyping sites for futureresearch Using a common set of genotypes across environments will alsoprovide the ability to monitor and detect emergence of new pathogenstrains that will be registered as shifts in disease pressure and emergingnew diseases, and how the environmental characteristics impacts pestbiology and prevalence CIMMYT has also developed several insect-pestresistant populations, inbred lines, and varieties, especially for the stemborers and postharvest insect-pests (weevils and grain borers) throughprojects such as Insect Resistant Maize for Africa (IRMA) In addition,several inbred lines have been developed combining resistance to stemborers and storage pests and these are currently being tested in easternAfrica Wide testing of these materials in Kenya, Tanzania, and Uganda isbeing done under IRMA

3.4 Breeding approaches for tolerance to

climate-related stresses

3.4.1 Conventional breeding

To increase the efficiency of breeding pipelines, a combination of tional, molecular, and transgenic breeding approaches will be needed.Breeding approaches are not mutually exclusive and are complimentaryunder most breeding schemes (Ribaut et al., 2010) Historically large gainshave been made through conventional breeding The success of the greenrevolution was based on breeding and resulted in large increases in cerealproduction (Eveson and Gollin, 2003) During the period of 1982–1994,the yield growth rate as a result of conventional breeding was 1.2%worldwide (Duvick and Cassman, 1999) In temperate maize, breedingbased on multi-location trials under different weather conditions hasresulted in increased grain yields at a rate of 73 kg
1 ha
1 yr
1 undermild stress (Duvick, 1997) In tropical maize, conventional breeding hasresulted in gains of up to 144 kg ha
1yr
1under drought stress (Edmeades

conven-et al., 1999) However, in the face of climate change, it is essential thatbreeding pipelines are improved to meet the needs of future generations Inconventional drought breeding, the application of proven breeding meth-odologies in managed stress screening has been attributed to the significantgains in grain yield under drought stress (Ba¨nziger et al., 2006) Up scalingtraining and application of these methodologies across projected drought-prone environments will play a key role in the continued development ofdrought adapted maize A similar approach will be required for additionalabiotic and biotic stress expected to increase under future climates.

A vast amount of research has focused on individual stresses However,

in the farmers’ fields the maize plants are regularly subjected to a tion of stresses Relatively little is known about the physiological andmolecular responses of crop plants subjected to stress combinations (e.g.,droughtþ heat or drought þ waterlogging); therefore, understanding theeffects of different individual stresses as well as their combinations is animportant step forward (Voesenek and Pierik, 2008) Breeding programsoften run independent screens for stresses know to occur in the targetenvironment, selecting genotypes which perform well across a suite ofstresses Independently screening for drought and low N tolerance intropical maize identified several physiological traits associated with toleranceunder one stress, conferred tolerance for the other stress (Ba¨nziger et al.,

combina-2000) Concurrent screening for both stresses successfully developed rior germplasm with tolerance to both stresses (Ba¨nziger et al., 2006).However, multiple stresses can have very different results and cannot bepredicted from the combination of individual stresses Mittler (2006).Rizhsky et al (2004) exposed the model specie Arabidopsis to heat anddrought stress simultaneously, and found that less than 10% of the regulatedgenes under combined heat and drought stress overlapped with the genesregulated by the individual stress treatments These findings implied that thegene networks that control different stress combinations cannot be reliablypredicted from those identified under specific individual stresses Predictedclimate change scenarios are likely to result in an increase in the stresses thatplants face in the field Given that combined tolerance to multiple stressesmay be different to individual tolerance, research needs to focus on stresscombinations likely to occur in the target environment This will beparticularly pertinent for drought stress and insect-pests combined, droughtand heat stress combined, and drought and waterlogging stress combined

supe-In the past 10 years, several institutions, especially in the private sector,have focused on the application of doubled haploid (DH) technology inbreeding programs, with an estimated 80% of companies employing thistechnology (Phillips, 2009; Ro¨ber et al., 2005) A DH is a genotype formedwhen haploid cells undergo chromosome doubling, allowing the produc-tion of a homozygous line after a single round of recombination.Blakelsee

et al (1922)reported the production of the first haploid plant, and the first

haploid maize was reported 10 years later (Randolph, 1932) The use of DHtechnology in breeding has the potential to increase the efficiency of linedevelopment by reducing the time taken to reach homozygosity in con-ventional breeding technology from approximately six seasons to one season(Mohan Jain et al., 1995) Initially, the efficiency of chromosome doublingmethods were too low for application within the maize breeding programs;however, Ro¨ber et al (2005) developed a temperate inducer maize linecalled RWS with a relatively high induction rate (8.1%), thereby increasingthe efficiency of DH development Tropically adapted maize inducer lineswith an induction rate of 10% are under development by CIMMYT, incollaboration with the University of Hohenheim (Prigge et al., 2011) Theability to apply DH technology within the tropical maize breeding couldsignificantly improve the genetic gains in the breeding programs Work iscurrently underway to transfer this technology to the African breedingprograms under the Bill and Melinda Gates Foundation funded project

“Drought Tolerant Maize for Africa”

Genetic diversity is an essential component of breeding progress; ever, to date, only a fraction of the available maize genetic diversity has beenutilized by the plant breeders Over 25,000 landraces, besides the wildrelatives teosinte and Tripsacum, 3000 elite inbreds, pools, and populations,are preserved in the CIMMYT Gene Bank (Ortiz et al., 2009) Within thegene pool of maize’s wild relatives, vast unexploited genetic diversity fornovel traits and alleles exists that could be used to broaden the genetic base ofbreeding and deliver beneficial genetic variation (Ortiz et al., 2009) Intensiveselection may have resulted in reduced genetic diversity for specific traits,either directly or indirectly Leveraging the hidden diversity within maizegene banks will potentially provide novel sources of favorable alleles tocomplement the ongoing breeding strategies While the landraces are notgenerally used directly by the plant breeders because of their poor agronomiccharacteristics, however, they can serve as sources of new inbred lines or DHlines from which new traits can be introduced into elite germplasm (Lafitte

how-et al., 1997) Simultaneously with the wider adoption of high-throughputmolecular tools, there is a distinct need to establish global phenotypingnetwork for comprehensive and efficient characterization of genetic resourcesand breeding materials for an array of target traits, particularly for biotic andabiotic stress tolerance and nutritional quality This would significantly accel-erate genomics-assisted breeding, diversification of the genetic base of elitebreeding materials, creation of novel varieties, and countering the effects ofglobal climate changes A new initiative coordinated by CIMMYT in collab-oration with many Mexican institutions, titled the “Seeds of Discovery” (SeeD),aims to discover the extent of allelic variation in the genetic resources ofmaize and wheat, formulate core sets based on genotyping and phenotyping,and utilize marker-assisted breeding to bring those rare useful alleles intobreeding programs for developing novel genotypes

3.4.2 Molecular breeding

The ability to quickly develop germplasm combining tolerance to severalcomplex polygenic inherited abiotic and biotic stresses will be critical to theresilience of cropping systems in the face of climate change Conventionalbreeding methods that rely on extensive phenotypic screening are effectivebut slow in producing germplasm tolerant to the current range of climaticconditions and are not optimal for rapidly improving tolerance to multiplestresses Molecular breeding offers the ability to increase the speed andefficiency of plant breeding (Whitford et al., 2010) Molecular breeding is

a general term used to describe modern breeding strategies where DNAmarkers are used as a substitute for phenotypic selection to accelerate therelease of improved germplasm Currently, the main molecular breedingschemes are MAS, marker-assisted backcrossing (MABC), marker-assistedrecurrent selection (MARS), and genome-wide selection (GWS), asdescribed inTable 3(Ribaut et al., 2010) Molecular marker-assisted breed-ing relies on the identification of DNA markers that have significantassociation with expression of specific target traits The use of moleculartechniques within breeding pipelines is widely and successfully employedwithin the private sector (Eathington et al., 2007) and with greater emphasis

in the public sector (Dwivedi et al., 2007; Whitford et al., 2010) Thedevelopment and availability of an array of molecular markers, greaterthroughput and reduced cost of genotyping assays, and above all, the recentavailability of the complete maize sequence within the public domain(Schnable et al., 2009) make the use of genotypic markers more accessiblewithin the public sector breeding programs Together these tools will allow

Table 3 Current molecular breeding strategies (adapted fromRibaut et al., 2010)

Genome-wide selection

(GWS)

Based on the prediction of performance Selection

is made on markers without significance testing and does not require the prior identification of markers associated with the trait of interest

key traits controlled by major genes as well as QTLs to be more efficientlyintroduced into breeding pipelines.

The application of molecular breeding requires identification of mic regions associated with the trait of interest Molecular markers, andmore recently high-throughput genome sequencing, provide the ability tocharacterize genetic diversity within the germplasm pool for most cropspecies (Moose and Mumm, 2008) Since the development of DNA markertechnology in the 1980s, great advances have been made in marker devel-opment, genetic maps, utilization of genome sequencing and the scale andcost of application of technologies (Dwivedi et al., 2007) QTL mapping hasbeen conducted for a wide range of traits, and extensive reviews have beenpublished on yield (Holland, 2007), biotic stresses (e.g., Balint-Kurti andJohal, 2009; McMullen et al., 2009; Wisser et al., 2006), abiotic stresses (e.g.,Collins et al., 2008; Salvi and Tuberosa, 2005; Wassom et al., 2008), anddomestication related traits (e.g., Doebley, 2006) Initial results suggestedplant populations generally segregate for a limited set of small effect QTLswith very few large effect QTLs (Salvi and Tuberosa, 2005) and QTLs werenot consistent across mapping populations Key factors likely to be respon-sible for these results are genetic heterogeneity and small mapping popula-tion sizes, resulting in skewed distributions of QTL effects (Beavis, 1998;Holland, 2007) However, several studies have now been published usinglarge population sizes for complex traits such as yield; while a large number

geno-of small effect QTLs were identified, together they accounted for less thanhalf of the total genetic variation (Schon et al., 2004) In general, a largenumber of small effect QTLs in maize have been identified for yield andabiotic stresses, while for many biotic stresses a few moderate to large effectQTLs have been identified

The identification of genomic regions associated with tolerance todrought stress has been the subject of much research in maize (Ribaut

et al., 2009) and other crops (for reviews, see Fleuery et al., 2010; Priceand Courtois, 1999) Drought studies have focused on the identification ofthe genetic basis of yield, yield components, and secondary traits includingincreased flowering synchrony (ASI), root architecture, growth mainte-nance, and stay green (see a review by Ribaut et al., 2009) A large QTLmapping study to identify stable genomic regions associated with yield,yield components, and flowering parameters identified over 1080 QTLs(Ribaut et al., 2009) Five QTL alleles for short ASI were introgressedthrough MABC from a drought-tolerant donor to an elite, drought-suscep-tible line Under severe drought, the selected lines clearly outyielded theunselected control However, their yield advantage decreased under mild tomoderate drought stress (Ribaut and Ragot, 2007) As suggested byCollins

et al (2008), the maintenance of biomass accumulation under water deficitshould be considered as an optimization process between transpiration,biomass accumulation, and its partitioning between root and shoot, rather

than as a tolerance process per se, and hence a given QTL can have positive,null, or negative additive effects depending on the drought scenario Thismay have considerably slowed the utilization of QTL data for breeding.Relatively less research has been conducted on the identification of QTLassociated with other abiotic stresses in maize, particularly for heat stress.Frova and Sari-Gorla (1994)identified QTLs associated with pollen toler-ance to a 2-h heat stress of 50 C during in vitro germination Using apopulation of 45 maize RILs, five QTLs associated with high-temperaturegermination and six QTLS for pollen tube growth were identified Veryfew overlapping regions for both traits were identified, implying that traitswere independently regulated Additionally no overlap was detected forQTLs under elevated and optimal temperatures A later study byFrova et al.(1998)using two maize mapping populations subjected to a heat stress (noinformation in terms of temperature and duration was provided) identifiedseveral QTLs associated with cell membrane stability, pollen germination,and pollen tube growth Using a larger mapping population in field condi-tions, a QTL accounting for 17% of phenotypic variation in grain yieldunder heat stress and 28% of the phenotypic variation in canopy tempera-ture on chromosome 4A was recently identified in wheat (Pinto et al.,

2010) In case of waterlogging tolerance in maize, several moderate effectQTLs have been identified for seedling stage tolerance to waterlogging (Qiu

et al., 2007) The authors screened a mapping population comprised on 288

F2:3 lines derived from a cross between tolerance (HZ32) and sensitive(K12) inbred lines under flooded (6 cm above the soil surface for 6 days)and nonflooded conditions in a series of pot experiments A total of 25 and

34 QTLs were identified in each experiment, accounting for between 4%and 37% of the genotypic variation in tolerance to flooding Moderate effectQTLs associated with shoot and root dry weight, total dry weight, plantheight, and a coefficient of tolerance for water tolerance were identifiedacross experiments on chromosomes 4 and 9.Mano et al (2005)developed

an F2mapping population between a maize inbred line (B64) and teosinte(Z mays ssp Huehuetenangensis) The mapping population was grown in apot experiment and flooded conditions were imposed for a period of

2 weeks QTLs associated with adventitious root formation under floodingcondition were identified on chromosomes 3, 7, and 8, Teosinte allelescontributed positively to all QTL confirming the potential use of Z maysssp Huehuetenangensis as a donor within breeding programs targeting water-logging tolerance A similar study using a different teosinte accession(Z mays spp Nicaraguensis) crossed to maize inbred line B73 identifiedQTLs controlling constitutive aerenchyma formation on chromosomes 1,

5, and 8 (Mano et al., 2009) The production of NILs containing theseQTLs from the donor Z mays spp nicaraguensis is underway and providing avalue genetic resource to confirm the potential of adventitious roots withaerenchyma to improved tolerance of maize to flooding

QTLs conferring resistance to major maize diseases (TLB, downy dews, SLB, rust, GLS, and many other diseases) and insect-pests have alsobeen identified (Balint-Kurti and Johal, 2009; Garcia-Lara et al., 2009;Krakowsky et al., 2004; Wisser et al., 2006) The first disease resistanceQTLs to be cloned in maize, Rcg1, for resistance to anthracnose stalk rot wasshown to be a Resistance Gene Analog (RGA) (Wolters et al., 2006) Anumber of mapping studies have been undertaken for all the major diseasesaffecting maize (see reviews byBalint-Kurti and Johal, 2009; Prasanna et al.,2010; Wisser et al., 2006) The disease QTLs mapping studies conductedthus far have provided information on the genetic architecture of diseaseresistance, including the number, location, and action of chromosomalsegments conditioning the trait Wisser et al (2006) further showed thatQTLs for resistance to different diseases often clustered together, mirroringthe clustered distribution of R genes and RGAs in plants A similar conceptwas proposed by McMullen and Simcox (1995) for disease and insectresistance related chromosomal regions in the maize genome There is aclear need for further genetic dissection of these QTLs rich chromosomalregions to more precisely localize the genes involved by developingQTL-NILs.

mil-The ultimate objective of QTL mapping is to identify the causal genes,

or even the causal sequence changes, known as quantitative trait nucleotides(QTN) (Holland, 2007) Initial QTL mapping only provides an approxi-mate localization to around 10–20 cM QTN identification requires a finermapping in a high resolution, detailed genetic complementation studies,and analyses of cosegregating sequence variants Fine mapping can be done

by selecting rare recombinants in the region of interest from very largepopulations that are nearly isogenic outside of the targeted region (Peleman

et al., 2005) With the large amounts of information available in publicdatabases like the whole genome sequence of B73 (www.maizesequence.org), and HapMapSNPs, maize is in an ideal setting for such fine-scalestudies At CIMMYT, work is currently underway for fine-mapping majorQTLs implicated in resistance to MSV, GLS, and northern corn leaf blight(NCLB) or TLB

Bernardo (2008)observed that when a large proportion of phenotypicvariation is controlled by many QTLs of small effects, the “find-and-intro-gress-QTL” approach has limited applicability due to overabundance ofQTLs identified for any given agronomic trait and their inconsistent effectsacross genetic backgrounds and environments Recurrent selection relies onthe phenotypic selection of superior progeny which are subsequentlycrossed with each other in every possible way to produce an improvedsource population thereby increasing the frequency of favorable alleleswithin a population With the rapid reduction in genotyping costs currentlyunderway, new genomic selection technologies have become available thatallow the breeding cycle to be greatly reduced and that facilitate the

inclusion of information on genetic effects for multiple stresses in selectiondecisions (Heffner et al., 2009) Three marker-based selection approachesare being utilized (F2enrichment, MARS, and GWS), that aim at increasingdesirable QTL allele frequencies in a population improvement context,either by utilizing the QTL information or without it, are increasinglygaining prominence.

Both the F2 enrichment and MARS (Bernardo, 2008) approachesrequire prior QTL identification through standard mapping procedures in

a suitable population and markers that are either linked to the QTLs orlocated within the QTLs In F2-enrichment, the individual F2 plants arescreened with informative markers and the unfavorable homozygotes areremoved to ensure all the remaining plants are carriers of desirable alleles(Bonnett et al., 2005; Wang et al., 2007) either in homozygous or hetero-zygous conditions This increases the probability of success of deriving asuperior recombinant inbred with smaller populations However, the effec-tiveness of this approach is reduced by the fact that only one generation ofmarker-based selection is performed in a typical F2 enrichment exercise,with an additional round in the latter stages also not being efficient MARSrelies on multiple rounds of marker-based selections with each cycle con-sisting of selected selfed progenies of each marker-selected individual andrecombining these progeny to form the next generation material, therebyovercoming the problems associated with F2 enrichment strategies F2

enrichment can target up to 9–12 unlinked QTLs MARS allows largernumber of marker loci to be targeted (up to 30); however, the products ofMARS (recombinant inbreds) may not be fixed for the favorable allele at alltarget loci (Bernardo, 2008)

GWS (often referred to as genomic selection) offers an alternativeapproach where no prior information on QTLs is required, with selectionbased entirely on the prediction of performance (Hamblin et al., 2011;Meuwissen et al., 2001) Genomic estimated breeding values (GEBVs) arecalculated for each individual in the population by fitting all the polymor-phic markers as random effects in a linear model and these are used for thebasis of selection Simulation studies using different numbers of QTLs (20,

40, and 100) and levels of heritability showed response to GWS was 18–43%higher than the corresponding responses using MARS (Bernardo and Yu,

2007) This suggests the potential of GWS for complex traits governed by alarge number of small effect QTLs Heffner et al (2009)suggested rapid-cycle genomic selection for abiotic stresses could increase genetic gains instress tolerance breeding by two- to threefold GWS has the potential tobypass problems associated with the number of QTLs controlling a trait, thedistribution of effects of QTL alleles, and epistatic effects due to geneticbackground (Bernardo and Yu, 2007), facilitating the inclusion of informa-tion on genetic effects for multiple stresses in selection decisions (Heffner

et al., 2009) New breeding and selection strategies like GWS rely on the

availability of cheap, robust, and reliable marker systems Pilot projects onthe implementation of rapid-cycling genomic selection using much highermarker densities are being initiated by CIMMYT on new platforms based

on next generation sequencing technologies, with the ultimate aim of itsroutine application across the CIMMYT and NARS maize breeding pro-grams in sub-Saharan Africa, Latin America, and Asia

3.4.3 Precision and high-throughput phenotyping

Breeding progress relies on genetic variability for the trait of interest (e.g.,grain yield under drought stress), high selection intensity through screening

a large number of genotypes and high broad-sense heritability for the trait ofinterest Improved phenotyping platforms will provide the foundation forthe success of conventional, molecular, and transgenic breeding Yield is afunction of many processes throughout the plant cycle thus integrative traitsthat encompass crop performance over time or organization level (i.e.,canopy level) will provide a better alternative to instantaneous measure-ments which only provide a snapshot of a given plant process (Araus et al.,

2008) Many new phenotyping tools based on remote sensing are nowavailable including nondestructive measurements of growth-related para-meters based on spectral reflectance (Marti et al., 2007) and infrared ther-mometry to estimate plant water status (Jones et al., 2009)

New phenotyping tools together with advances in molecular technologieswill be a powerful combination toward rapid advances in germplasm improve-ment However, to ensure the full potential of such tools, greater emphasisneeds to be given to reducing the within-experimental site variability Fieldsexperiments provide the cornerstone for all germplasm development; how-ever, the importance of environmental uniformity and good agronomicpractices are often overlooked Without uniform phenotyping field sites themuch anticipated benefits of molecular breeding will not be realized Highlyvariable field sites will produce highly variable data, thereby masking impor-tant genetic variation for key traits, regardless of the cost and precision of aspecific phenotyping tool Phenotypic variation among individuals could bedue to genetic and environmental factors Broad-sense heritability estimates,therefore, reflect the amount of variation in genotypic effects compared tovariation in environmental effects Heritability is specific to a specific popula-tion within in a specific environment and can be reduced due to increasedenvironmental variation without any genetic change occurring Broad-senseheritability (H) is defined as the proportion of phenotypic variation that is due

to genetic variation (Falconer and Mackay, 1996) and is defined as

s2Gþs 2 GE

e þs 2 E re

G is the genotypic variance ands2

E is the phenotypic variance, e isthe number of environments or locations, and r is the number of reps,calculated from variance components obtained from an analysis of variance.Since phenotypic variation of a population is caused by both genetic (signal)and environmental factors (noise), broad-sense heritability provides a usefulestimate to determine the proportion of phenotypic variance that can beattributed to genetic effects Broad-sense heritability is population specificwithin a particular environment and typically decreases with increasedsite (environmental) variability As a result, by identifying and implement-ing methods to reduce environmental variation within agricultural trials,broad-sense heritability can be increased resulting in potentially greaterselection gains

Increasing trial heritabilities through reduced environmental error istherefore essential to improve the cost-effectiveness of phenotyping andincrease the genetic progress in the development of climate-ready germ-plasm This is particularly pertinent for breeding for abiotic stress tolerance,where variability can be masked under optimal conditions (Ba¨nziger et al.,

2000) Soil variability is a major cause of inherent site variability Additionalgenerators of within-site variability include topography, bordering, andcrop management (Blum, 2011) A recent review of field variability withinrice drought phenotyping sites highlighted variability in soil physical prop-erties within and between experimental sites (Cairns et al., 2011) In generalrelatively little is known, and even less reported, about soil properties andvariability within phenotyping Initial characterization of field sites prior touse for phenotyping will allow researchers to exclude sites where largeexperimental error is likely to be introduced through highly variable soilproperties In phenotyping sites with moderate to high heterogeneity,variability maps of important characteristics for specific trials (e.g., soiltexture for drought trials and residual nitrogen levels for low nitrogen trials)will allow researchers to avoid areas of high variability or design trialsincorporating spatial variability Experiments can be planted within areas

of the least spatial variability and/or individual trials blocked within bility gradients to reduce within experiment or within replicate environ-mental error (Cairns et al., 2004, 2009)

varia-Site characterization is often used for precision agriculture applicationsbut is less frequently applied within public breeding programs Manytechniques are available for mapping variability within field sites based onsoil sampling, soil sensors, and measurements of plant growth as surrogates

of variability Destructive soil sampling for key soil physical and chemicalproperties conducted on a grid sample can provide a low-cost measure ofsoil variability Soil texture strongly influences water holding capacity,water release characteristics, and nitrogen mineralization (Marshall et al.,

1996) Within site variability in soil texture can introduce variation in thedevelopment of drought stress as a result of the variation in water release