Intended and unintended consequences of genetically modified crops – myth, fact and/or manageable outcomes?

ABSTRACT

Food and feed has been produced from genetically modified (GM) crops for 25 years. It is timely to review whether this technology has globally delivered the expected benefits and whether the ongoing debate on risks is justified. Expected benefits associated with GM include increased crop yields, reduced pesticide and insecticide use, reduced carbon dioxide emissions, improved soil structure, improved crop nutritive quality/value, and decreased costs of production. Concerns focus on food safety linked to toxicity and allergenicity, environmental risks associated with potential chances of gene flow, adverse effects on non-target organisms, evolution of resistance in weeds and insects, and genetic perturbations resulting in unintended compounds, new diseases, or antibiotic resistance. This review focusing on benefits and risks of GM crops concludes that they are a valuable option for delivering improved economic and environmental outcomes by providing solutions for many of the challenges facing mankind. GM technologies like many non-GM technologies can bring risks, but these can and have been monitored and quantified, allowing decisions balancing commercial, societal and environmental benefits against measurable risks. While ‘checks’ and ‘balances’ are required, regulatory schemes must focus on balancing risks and benefits and not on ‘checks’ alone which is the case for many countries.

KEYWORDS:

• Biotechnology
• environment
• food safety
• gene editing
• GMOs
• transgenic

Introduction

Genetic modification resulting in altered DNA nucleotide sequences is an overarching term which covers several different methods used in manipulating the genome of organisms. Genetic modification does not refer to breeding or mutagenesis, which can also modify genetic sequences but does include transgenic manipulation, gene stacking, targeted gene editing using site-directed nucleases resulting in gene deletion, modification or gene insertion, and RNA gene silencing (Halpin 2005; Barrangou et al. 2007; Douglas and Halpin 2010; Kuzma and Kokotovich 2011; Shan et al. 2013; Arujanan and Aldemita 2015; Rinaldo and Ayliffe 2015; Schwach and Baulcombe 2016; Mall et al. 2018; El-Mounadi et al. 2020; ISAAA 2021a; Rahman et al. 2022). It can be a complex and costly process (Prado et al. 2014). The ‘Biosafety Clearing House’ managed by the Convention on Biological Diversity records 913 living modified organisms developed for resistance to diseases and pests, resistance to herbicides and antibiotics, tolerance to abiotic stresses, changes in physiology and or production, changes in quality and/or metabolite content, production of medical or pharmaceutical compounds, use in industrial applications, and engineered gene drive application (Convention on Biological Diversity 2022). A significant driver for the implementation of GM technologies is the ‘genetic glass ceiling’ within species, where the genes available for traits are not available within the species, that classical breeding cannot overcome (Gressel 2010).

Genetically modified (GM) food and feed crops have been acknowledged as a technology that has been adopted at a faster rate than any other recent crop technology (James 2010; Adenle 2011; Khush 2012; Flavell 2015; James 2015; Kamle et al. 2017; Raman 2017; ISAAA 2018a; Nwosu and Ubaoji 2020; Prabha et al. 2020; Scheitrum et al. 2020). There have been 525 different transgenic events undertaken in 32 crops and flower species that have been approved for cultivation in different parts of the world (Kumar et al. 2020; Verma et al. 2021) now used on 192 million ha in 2018 (ISAAA 2018a). Between 1992 and 2018 regulatory agencies across 43 countries, including New Zealand, (with the European Union counted as one country) have approved 2063 GM foods and 1461 GM feeds by considering them to be as safe as non-GM crops (ISAAA 2018b). This is underpinned by 824 approvals allowing the cultivation of GM crops (not including approvals for ornamental crops such as carnation, rose, and petunia) for eight types of GM traits (Table 1) (ISAAA 2022). However, the predominant commercialised GM crops are those providing glyphosate tolerance to improve weed control (Green and Owen 2011; ISAAA 2017) and those incorporating genes expressing insecticidal proteins from Bacillus thuringiensis (Bt) (Betz et al. 2000; Shelton et al. 2002; Duke and Powles 2009; Barrows et al. 2014a) (Table 1). For fruit and vegetables, post-harvest management to improve shelf life and quality is equally important and could be achieved using genome editing tools (Kumari et al. 2022).

Table 1. Number of commercial GM events by trait and crop species – summarised from ISAAA (2022).

Crop species	GM trait
	Abiotic stress tolerance	Altered growth/ yield	Disease resistance	Herbicide tolerance	Insect resistance	Modified product quality	Nematode tolerance	Pollination control system
Apple						3
Argentine canola				37		10		24
Bean			1
Chicory				3				3
Cotton				45	50	1
Cowpea					1
Creeping bentgrass				1
Eggplant					1
Eucalyptus		1
Flax				1
Lucerne				4		2
Maize	7	2		215	210	14		6
Melon						2
Papaya			4
Pineapple						1
Plum			1
Polish canola				4
Poplar					30
Potato			19	4	8	18
Rice				3	3	2
Safflower						2
Soybean	2	1		35	6	12	1
Squash			2
Sugar beet				3
Sugarcane	3
Sweet pepper			1
Tobacco				1		1
Tomato			1		1	9
Wheat	1			1
TOTAL	13	4	29	357	310	77	1	33

Currently, GM crops are grown in 26 (ISAAA 2018a) countries with the top 5 ranked by area grown being – USA, Brazil, Argentina (Mühl 2020), India, and Canada (Kamle et al. 2017). In 2018, 21 developing countries grew 54% of the global GM crops by area (ISAAA 2018a). In 2018, a total of 70 countries (26 allowing planting and 44 not allowing planting of GM crops) adopted 30 GM crops for food, feed, and cultivation (ISAAA 2018b). The range of plant species grown as commercial GM crops is diverse including row crops, vegetables, and fruit trees (Parisi et al. 2016) but the predominant crops are maize (Zea mays), soybean (Glycine max), cotton (Gossypium hirsutum), and canola (Brassica napus) (James 2016; ISAAA 2018a), many of which have multiple stacked traits (Parisi et al. 2016; Shehryar et al. 2020). It is interesting that wheat (Triticum aestivum), a major food and feed crop has only one commercialised GM cultivar (Table 1) despite the claims nearly 20 years ago that gene transformation will be an important tool for developing cultivars for abiotic and biotic stress tolerance and the value-added characteristics to improve human health (Bhalla 2006). This was the result of economic concerns from wheat growers and the processing industry using the same argument that delayed using GM potatoes, sugar beet and rice (Kalaitzandonakes et al. 2016).

Despite this unprecedented uptake there is still the ongoing debate on the balance of benefits and risks associated with this technology (Beringer 2000; Griffin 2000; Bakshi 2003; Conner et al. 2003; Hunt 2004; Cerdeira and Duke 2006; Parrot 2010; Buiatti et al. 2013; Kramkowska et al. 2013; Goldstein 2014; Zhang et al. 2016; Smyth 2017a; Kumar et al. 2020; Kavi Kishor et al. 2021). Political interference has also significantly impacted regulatory approval processes of GM crops adversely affecting the adoption of innovative, yield enhancing crop varieties, thereby limiting food security opportunities in food insecure economies (Davison 2010; Phillipson and Smyth 2016; Raybould 2021; Smyth et al. 2021). Herring (2010) has considered that unlike some global contentions the rift over GM crops is not about ultimate values but rather about bio-safety and bio-property. In the case of bio-property, associated with patents, monopoly corporate control and terminator technology, claims have proved either false or inconsistent with dynamics on the ground.

One of the commonly stated justifications for pursuing GM crops has been the need to feed a growing world population (Huang, Pray, et al. 2002; Christou and Twyman 2004; von Braun 2010; Adenle 2011; Tait and Barker 2011; Rezbova and Skubna 2012; Okeno et al. 2013; Reddy et al. 2013; Whitty et al. 2013; Francis et al. 2016; Georges and Ray 2017; Oluwambe and Oludaunsi 2017; Taheri et al. 2017; Chaudhuri and Datta 2018; Tyczewska et al. 2018; Ehirium et al. 2020; Nalluri and Karri 2020; Ahmad et al. 2021; Girish et al. 2021; Guleria and Kumar 2021; Sood et al. 2021; Verma et al. 2021; ISAAA 2021a, 2021b). While this is certainly a potential outcome and for many scientists is a significant driver, there are other underlying reasons for companies and research groups pursuing the use of GM crops which include commercial returns and scientific recognition, neither of which are necessarily wrong but must be recognised as material.

Benefits associated with the use of GM have included increased crop yields which should lead to a reduced need for new land for production, reduced pesticide and insecticide use, reduced carbon dioxide emissions, improved soil structure, a decreased cost of crop production, and health benefits through provision of improved micronutrient supply (Barrows et al. 2014b; Kumar et al. 2020; Wu et al. 2021), many of which are associated with adapting to or mitigating predicted outcomes from climate change (Karavolias et al. 2021). Indeed, GM is another tool that can deliver benefit primarily with regard to improved environmental integrity and increased productivity (yield per ha), through reducing the use of synthetic chemicals (which no doubt cause environmental harm and possibly health concerns (Devine and Furlong 2007; Boedeker et al. 2020; Fu et al. 2022; Rasool et al. 2022; Sim et al. 2022)) and reducing the need for more land to be taken out of natural ecosystems for production agriculture. Boedeker et al. (2020) have estimated that globally there are annually about 385 million cases of unintentional, acute poisoning due to synthetic pesticide use annually including about 11,000 fatalities, with worst affected areas being southern Asia, followed by south-eastern Asia and east Africa. However, the world continues to use pesticides. Can genetic modification technologies provide a safer and more environmentally benign option?

Concerns about the use of GM technologies (Kendall et al. 1997) have included:

1. food safety linked to potential toxicity and allergenicity,

2. environmental risks associated with potential chances of gene flow, adverse effects on non-target organisms, and evolution of resistance in weeds and insects, and

3. genetic perturbations leading to the production of unintended compounds, new disease, or antibiotic resistance.

However, The Royal Society London and National Academies of Science, Engineering and Medicine (NASEM) concluded after two decades of GM crop use that ‘there is no evidence that producing a new crop variety using GM techniques is more likely to have unforeseen effects than producing one using conventional cross breeding’ (NASEM 2016; The Royal Society 2016). Yet unintended consequences of any new crop cultivar (GM or non-GM) needs risk assessment, which is regulated for GM crops in most jurisdictions (Key et al. 2008; DeFrancesco 2013; Kumar et al. 2022), and strongly regulated in New Zealand (Caradus 2022) but is not regulated for non-GM crops.

Not all GM crop developments will be successful, and many will end up being ‘shelved’ for a variety of reasons. For example, the use of a transgene trypsin proteinase inhibitor to enhance pest resistance did not show increased mortality, but rather a net gain in weight and a faster development compared with control larvae (De Leo et al. 1998). However, failed developments also occur when using non-GM options when attempting to improve crop adaptation and yield (Mehta et al. 2019). Similarly, poor product stewardship can result in otherwise successful GM crop developments failing, for example as has happened with Starlink maize and a herbicide-tolerant wheat (Mbabazi et al. 2021).

This systematic review will focus on plants (but not animals or microbes) and will use peer-reviewed sources to determine if intended consequences of GM technologies are being delivered and whether unintended consequences are real and if so, are they manageable? This will cover assessments made to determine the risks and opportunities of using GM crops (e.g. refer Figure 2 of Carzoli et al. 2018). The purported or expected benefits of commercialised crops with GM traits will be reviewed followed by an assessment of unintended consequences, what they mean and how they might be managed. The influence of corporations and governments on the uptake and attitudes towards GM crops will also be briefly reviewed. The review will not include either unregulated techniques used to manipulate the genome or issues associated with globalisation, regulatory frameworks and intellectual property associated with GM crops, consumer attitudes or cultural (Young and Cormick 2004), ethical and religious concerns, or use in medical or industrial molecular farming (Daniell et al. 2001). However, it is acknowledged that the debate on the use of GM technologies is simply not one based on science pros and cons but has significant social, economic, and cultural overlays (Cayford 2004). However, these are largely beyond the scope of this review.

There are 7 sections to this review: (1) a brief commentary on the influence of corporations and governments; (2) expected benefits from GM crop; (3) GM product quality and safety testing for consumption by either animals or humans; (4) unintended consequences of traits in GM crops – fact or myth; (5) coexistence of GM and non-GM crops – is it possible or does it create more issues; (6) future opportunities for GM crops; and (7) a concluding commentary.

Influence of corporations and governments

GM crops have been promoted and championed by some key agricultural corporations (Macnaghten 2015; Thuy 2018; Prabha et al. 2020). It has been argued that because of some large agricultural companies, transgenic seeds are too expensive for poorer farmers (ETC Group 2013). It has been propositioned that the inclusion of genetically modified maize in food aid shipments to Southern Africa during the 2002 food crisis was done to promote the adoption of biotech crops in the continent, expanding access and control of multinational companies and as a result undermining local smallholder production thereby fostering greater food insecurity (Zerbe 2004). On the other hand, in other countries, such as China, there has been a move by government researchers to promote the use of GM plants (Cao and Li 2013), particularly Bt GM cotton, amongst small holder farmers (Huang, Hu, et al. 2002; Hong et al. 2013). However, confidence in how a government is seen to be handling the management of GM crops will impact on support for the commercialisation of these technologies (Yu et al. 2020).

Additionally, both government bureaucratic regulatory systems, attitudes of corporations and influence of antagonist organisations can slow or prevent the delivery of public good GM technologies. This has occurred in New Zealand (Caradus 2022) and Europe where most of the European Union farmers are unable to grow GM crops (Zilberman et al. 2013; Lucht 2015). Another excellent example of the influence of government and political economy is the deployment struggles associated with the release of Golden Rice (Oryza sativa) (Ye et al. 2000; Beyer 2010; Potrykus 2010; Wesseler and Zilberman 2014; Wu et al. 2021), which is genetically modified to elevate expression of pro-vitamin A, a deficiency which can cause blindness and reduced life expectancy (Sommer and West 1996). Early criticism came from organisations against GM crops by indicating that the levels of vitamin A produced in the early transformations were inadequate to make a real difference (Ensering 2008). However, this has been corrected with Golden Rice now documented to contain up to 35 μg β-carotene per gram, which can then be effectively converted to vitamin A in humans (Tang et al. 2009). Golden Rice is now used commercially in Philippines (Sumangil 2022) but should have a much wider use in populations suffering from vitamin A deficiency (Anderson et al. 2005; Stein et al. 2006). An interesting thought piece is provided by Stone and Glover (2017) on the prospects for Golden Rice in the Philippines prior to its commercial release through an analysis of three distinctive ‘rice worlds’. Green Revolution rice developed at the International Rice Research Institute (IRRI) 60 years ago, Golden Rice being bred at IRRI in the early 2000s, and a scheme to promote and export traditional ‘heirloom’ landrace rice from the Philippines.

Alternatively, political support within a country for the use of GM crops can result in laws and regulations being passed that provide a food regime based on biotechnology as occurred in Brazil late in the first decade of 2000, despite vehement opposition (Motta 2016). The motivations and drivers for governmental decision of this kind often trace back to agrarian elites, multinational seed and chemical corporations, and national research institutes (Murrell 2011). The varied uptake of GM technologies across Africa was largely driven by strong governmental support for GM technology (Schroeder 2022) or a lack of political support and ‘will’ to embrace it (Okeno et al. 2013).

Despite the calculated increased profit from Bt GM cotton of US$150/ha compared to US$70/ha for non-GM cotton (Vitale et al. 2014), Burkina Faso announced an abrupt phase-out of Bt cotton, citing millions of dollars of losses due to inferior lint quality (Luna and Dowd-Uribe 2020). They concluded that it was not the GM process that was used to develop this technology that was at fault here, but rather the political economic context that favoured the production of positive results, the narrow epistemologies of agronomic evaluation, and the knowledge produced via these processes was used to accrue material benefits to the IP owner and helped to promote GM crops across the continent. They stress that future GM crop evaluations should be more self-reflexive, critical, and transparent in how power shapes the evaluation process and agricultural outcomes for farmers. Indeed some governments in Africa have taken advice to adopt the European regulatory approached and driven GM food and crops from their economies, this despite the fact that in Africa two-thirds of farmers are poor and in desperate need of new technologies to boost crop productivity (Paarlberg 2009).

Concern for farmers being held hostage to the ‘agro-industrial machine’ and therefore the use of GM technologies (Tirado and Johnston 2010) which while unacceptable is not a fair basis for condemning GM technologies per se. Other issues affecting equitable uptake of GM technologies include government credit provisioning schemes because of high seed prices (Dowd-Uribe 2014); suspicion of direct or indirect financial support for science teams promoting GM crop technologies (Kangmennaang et al. 2016); the argument that GM technologies are needed to secure future food production is simply a reflection of corporate interests (Jacobsen et al. 2013); and concern about corporate ownership of seed and threats to the purity of indigenous crops (Altieri 2003; Schmidt 2005). The domination of large multinationals in the development of GM crops while concerning should not be specifically argued as the reason for condemning GM technologies and the benefits they can provide. The focus should rather be on condemning poor business practices and a lack of humanitarian concern. It cannot necessarily be assumed that improved productivity will result in improved food security (Fukuda-Parr and Orr 2012) unless there are better (fairer) means for food distribution. Geopolitical and corporate power plays should not be a reason for condemning the technology when there are significant advantages on offer. Indeed, a European Parliament study using commonly applied metrices for measuring power has revealed that there is no monopolistic pricing power present (Wesseler et al. 2015).

Genetically modified crops – expected benefits from traits

While the first GM crop (GM tobacco for virus resistance) was commercialised in China in 1990 (Raman 2017) followed by the Flavr Savr tomato (Solanum lycopersicum) (Kramer and Redenbaugh 1994; Bruening and Lyons 2000; Martineau 2001), early traits commercialised were dominated by input traits such as glyphosate herbicide tolerance and Bt pest resistance (Huang et al. 2005; Qaim 2015). These and additional traits now available in GM crops along with their expected benefits are listed in Table 2.

Table 2. Commercialised GM traits and their benefits – (also refer to Halford 2004; Nalluri and Karri 2020; Brookes and Barfoot 2020c for more detail).

Trait	Crop	% Total GM crops*	Intended consequence; efficacy and benefits	Reference
Glyphosate tolerance	Range of crops – maize, cotton, canola	–	Reduced use of a broad range of selective herbicides; replaced with used of broad spectrum herbicides such as glyphosate	Brookes and Barfoot (2020c)
	Canola	3.8	Weed control	Monjardino et al. (2005)
	Cotton	2.3	Improved weed control; reduced tillage	Ammann (2005); Dill et al. (2008)
	Maize	20.5	Weed control	Dill et al. (2008)
	Potato (Solanum tuberosum)	<0.1	Weed control	Bakhsh et al. (2020)
	Soybean	35.5	Improved weed control; reduced tillage	Ammann (2005); Dill et al. (2008)
	Sugar beet (Beta vulgaris)	0.2	Economic; weed control	Demont and Dillen (2008)
	Wheat	<0.1	Wheat transgenic event 33391 gave good glyphosate tolerance and yield	Zhou et al. (2003)
Insect pest resistance due to Bt Cry proteins	Range of crops – maize, cotton	–	Significantly reduced insecticide use	Betz et al. (2000); Brookes and Barfoot (2020c)
			30 pests controlled by Bt crops in Latin America	Blanco et al. (2016)
			Bt crops suppress pests, reduce pesticide use and increase farmer profits	Tabashnik and Carriere (2015)
	Egg plant, Brinjal (Solanum melongena)	<0.1	Control of major pests	Muhammad et al. (2020)
	Cotton	8.7	Control of cotton bollworm, Helicoverpa armigera	Lu et al. (2012); Jadhav et al. (2020)
	Cowpea (Vigna unguiculata)	<0.1	Control of bean pod borer, Maruca vitrata, which can cause 20–80% yield loss	Addae et al. (2020)
	Maize	19.3	Control of the European maize borer	Hutchison et al. (2010); Chvatalova (2020)
			Control of maize rootworms (Diabrotica spp.) – meta analysis of 10 publications	Pellegrino et al. (2018)
	Potatoes for industrial use and animal feed	<0.1	Control of Colorado potato beetle	James (2011); Kadoić Balaško et al. (2020)
	Rice	<0.1	Control of yellow stem borer (Scirpophaga incertulas), rice striped stem borer (Chilo suppressalis), pink stem borer (Sesamia inferens), and rice leaf roller (Cnaphalocrocis medinalis)	Bashir et al. (2004); Ramesh et al. (2004); Tang et al. (2006); Li et al. (2016)
	Soybean	9.8	Control of key soybean lepidopteran pests	Grossi-de-Sa et al. (2011); Bacalhau et al. (2020)
Combined herbicide tolerance, male sterility induction and insect resistance	Maize	<0.1	MON 87427 × MON 89034 × NK603 developed by conventional breeding	Venkatesh et al. (2014)
Disease resistance	Papaya (Carica papaya)	<0.1	Resistance to papaya ringspot virus	Lius et al. (1997); Gonsalves (1998); Gonsalves et al. (2007)
	Rice	<0.1	Resistance to bacterial blight (Xanthomonas oryzae)	Rani et al. (2008)
	Squash (Cucurbita sp.)	<0.1	Virus resistance	Fuchs and Gonsalves (1995); Tricoll et al. (1995)
Reduced lignin	Alfalfa (lucerne) (Medicago sativa)	<0.1	Improved forage digestibility	ISAAA (2017); Barros et al. (2019)
Drought tolerance	Maize	<0.1	Increase grain production in high-water-deficit environments	Chang et al. (2014)
			Use of event MON 87460 to improve drought tolerance though improved water use efficiency	ISAAA (2016)
Phytase expression	Maize	<0.1	Replace the additional P supplementation in monogastric animals, reducing the cost of feeding	Wang and Guan (2020)
Improved nutritional quality	Rice	<0.1	Expression of Vitamin A (Golden Rice); many years after development has become commercial in Philippines	Paine et al. (2005); Tang et al. (2009); Kim et al. (2010); Stone and Glover (2017); Sumangil (2022)

*Percentage of GM crop plantings in 2018 – from Figure 11 of Brookes and Barfoot (2020c); Those categorised as <0.1% were not included in the Brookes and Barfoot (2020c) review – so this level is below the lowest quoted by them.

In 2014, a meta-analysis indicated there were significant benefits of GM crops on yield, pesticide use and cost, production costs and farmer profit (Table 3) (Klümper and Qaim 2014). The number of different primary datasets used for this analysis ranged from 7 (for pesticide use associated with herbicide-tolerant crops) to 100 for yield across all GM crops. The analysis showed no significant difference in farmer profit (+64%) with the use of herbicide-tolerant crops. However, the primary reason for this was that there were only 9 primary datasets, which were quite variable, used. Likewise, studies comparing conventional systems with those focused on soil and water conservation and/or organic farming have shown mixed results in terms of economic productivity, due to uncertainty of higher price premium and government support to offset lower yields and higher overall economic costs (OECD 2016).

Table 3. Calculated average percentage differences between GM and non-GM crops for important output variables from a meta-analysis by Klümper and Qaim (2014).

Output variable	Type of GM crop
	All GM crops	Insect resistant	Herbicide tolerant
Yield	+21**	+25**	+9*
Pesticide quantity used	−37**	−42**	+2
Pesticide cost	−39**	−43**	−25**
Total cost of production	+3	+5*	−7
Farmer profit	+68**	+69**	+64

*p < 0.05; **p < 0.01.

Economic gains

The economic benefits globally from GM crops between 1996 and 2016 have been estimated to be close to US$200 billion, benefiting up to 17 million farmers of which 95% are from developing countries (ISAAA 2018b). In 2018 the on-farm income benefit from the use of GM crops was estimated to be close to US$19 billion, equivalent to adding over 5.8% to the global value of the crops soybean, maize, canola, and cotton (Brookes and Barfoot 2020a). For the 22-year period to 2018, the net economic benefit at the farm level has been close to US$225 billion with these gains almost equally divided between farmers in developed and developing countries (Brookes and Barfoot 2020b). In the following 2 years, through to 2020, a further US$36.3 billion has been added to this economic benefit with 72% of the gains derived from yield and production gains and the remaining 28% coming from cost savings (Brookes 2022). A further independent estimate has determined that at the current farm-level GM adoption rates, the increased production achievable could generate an additional $65 billion, the majority of which would accrue to the developing world (Scheitrum et al. 2020). Interestingly, early on in commercial development the likely economic benefit of glyphosate-tolerant soybean in USA was examined with the view that the price premium was being set too high compared to the potential cost/risk savings on many farms, and as a result ‘Roundup Ready soybean will not be fully adopted soon’ (Bullock and Nitsi 2001). How wrong that prediction has proven, with glyphosate-tolerant GM soybean being the most cultivated transgenic plant in the world in 2006, and in the USA, making up 91% of soybean crop in 2007 (Bonny 2008), with an estimated cumulative global benefit over 15 years to 2010 of US$46 billion (Alston et al. 2014). The uptake of this technology was due to convenience and ease of use (Marra et al. 2004), a major driver for the successful uptake of other technologies such a glyphosate in Africa (Haggblade et al. 2017) and Epichloë endophytes in New Zealand (Caradus et al. 2013)

Of the 98 results from peer-reviewed literature that compare the economic performance of GM crops to their conventional counterparts, 71 indicate a positive impact, 11 neutral and 16 negative (Carpenter 2010). For GM herbicide-tolerant crops, 12 of 17 results show a positive impact on economic performance, whereas 4 results show no difference and 1 result shows a negative impact. For GM insect-resistant crops, 59 of 80 results indicate improved economic performance, 7 results are neutral, and 14 results are negative. In Western Canada farmers growing herbicide-tolerant GM canola generated between CAD$1.063 billion and CAD$1.192 billion annual net direct and indirect benefits from 2005 to 2007, partly attributed to lower input costs and partly attributed to better weed control (Gusta et al. 2011). In Western Australia, the economic value of using glyphosate-tolerant GM canola in the management of herbicide-tolerant annual ryegrass (Lolium rigidum) and wild radish (Raphanus raphanistrum) was considerably higher than the commonly grown triazine-tolerant canola (Monjardino et al. 2005) which also caused a yield and oil content penalty compared with cultivars not tolerant to triazine (Robertson et al. 2002). In Australia, the economic opportunity cost from 2004 to 2014 resulting from a moratorium on GM crop use resulted in a net economic loss to canola farmers’ of AU$485.6 million (1.1. million tonnes) (Biden et al. 2018). While the initial uptake of herbicide-tolerant GM canola was lower than expected in Australia due largely to economic value derived from the GM canola technology package (Hudson and Richards 2014), it is now more widely used reaching half million ha in 2018 (Brookes and Barfoot 2020b) out of a total of about two million ha (ABARES 2018). For the Czech Republic and Hungary, an estimated economic analysis showed that the use of herbicide-tolerant GM sugar beet would provide the farmer additional return of €252–262/ha of which two-thirds accrue to farmers and one-third to the seed industry (seed company and technology provider) (Demont and Dillen 2008). Similar outcomes have been reported for GM sugar beets (Demont et al. 2004) and GM maize (Wesseler et al. 2007). The concerns about GM maize imports into Mexico (Hernández-López 2022) and the recent announcement that Mexico is to ban GM maize will have a significant economic impact both in Mexico and the USA. It is forecast that over the next 10 years this Mexican ban will cause the USA economy to lose US$73.89 billion in economic output, and for Mexico a reduction of US$19.39 billion (World Perspectives Inc 2022).

The economics of delaying the introduction of GM crops due to concerns about their unintended effects (i.e. unexpected environmental side effects) has shown that globally restriction of the adoption of GM in corn, rice, and wheat can only be justified if the net present value of the damage caused is above US$300 billion to US$1.22 trillion (which is the estimated value of the benefit using these technologies), suggesting that a precautionary approach can be very costly (Zilberman et al. 2015). In the USA, a 162 page report on the use of GM and non-GM feed concluded that demand for use of non-GM feed only for livestock and poultry would most likely increase greenhouse gas emissions on farms, and raise consumer prices for meat, milk, and eggs (Bowers 2022). However, early in their use the positive influence of some GM crops on economic competitiveness in some parts of the USA has been questioned (McBride and El-Osta 2002; Jolly et al. 2005). Additionally, the uptake and use of GM soybean in Argentina while cited regularly as an economic success through providing increased production and profits (Trigo and Cap 2003; Qaim 2005; Qaim and Traxler 2005; Chudnovsky 2006) has also been criticised for creating profound changes in social and ecological dynamics affecting equality, displacement, health hazards, deforestation, and creation of glyphosate-resistant weeds suggesting that the model was unsustainable (Altieri and Pengue 2006; Leguizamón 2014; Phélinas and Choumert 2017). However, nearly a decade later Argentina remains a major producer of glyphosate-tolerant GM soybean involving 17 million ha in 2018 (Brookes and Barfoot 2020b), much of which is exported to China (Giraudo 2020).

The cost to farmers, particularly small-holder farms in developing countries, of using GM crop technologies is often used as a criticism (Raney 2006; Moscona and Sastry 2022). In fact a review showed that the economic evidence does not support the widely held perception that transgenic crops benefit only large farms in developed countries (Lee et al. 2014) but on the contrary, the technology may also be a benefit to poorer small farms (Huang, Rozelle, et al. 2002; Ruttan 2004; Bennett et al. 2004a; Huang et al. 2005). Comparing the benefit to farmers in developing countries with those from developed countries indicates a 20% and 36% cost of use relative to the total returns from the technology, respectively (Brookes and Barfoot 2017). This difference is partly due to both weaker intellectual property rights enforcement in developing countries (Krattiger 2010) and the higher average level of farm income gain on a per hectare basis derived by developing country farmers relative to developed country farmers. In a study undertaken in Uganda where there are farmer estimated yield losses in maize to drought and stem borer of 36% and 15%, respectively it was found that using GM maize with increased drought tolerance and/or Bt insect resistance provided a value/cost ratio of 1.5 and 2.1, respectively (Wamatsembe et al. 2017). The only situation where a negative return occurred was for farmers with non-stressed grain that was yielding below 2 t·ha⁻¹. This is supported by an earlier review which only used peer-reviewed literature with a stated method applied to empirical data to show that the use of GM crops by developing economies is more than not likely to be economically beneficial (Smale et al. 2009); and from economic analyses of impact studies (Qaim 2009). Indeed, in Africa urban rich consumers rejected GM-banana while poor urban and poor rural consumers and producers showed a positive willingness to purchase (Kikulwe et al. 2011) confirming the hypotheses of Paarlberg (2009) and Thomson (2013) that the better educated wealthy urban consumers orient themselves more towards the views expressed by wealthy EU consumers than by poor African consumers.

Improved productivity and crop yield performance

There is a need to consider how more can be produced from current arable and grazed pasture-land without further destruction of forests and wetlands (Ewers et al. 2009; Ray et al. 2012; Barrows et al. 2014b; Phalan et al. 2016; Folberth et al. 2020). It has been estimated that crop production globally has increased by a cumulative 658 million tons over non-GM equivalents over the 20 years from 1996 to 2016 and in so doing has reduced the area requiring to be cropped by 183 million ha (ISAAA 2018b).

Yields of GM crops of soybean, maize, canola, and cotton have increased globally since 1996 compared with equivalent non-GM crops (Brookes and Barfoot 2017). Using aggregate data from across-country time series GM crops were estimated to have increased yields by 34% for cotton, 12% for maize and 3% for soybeans (Barrows et al. 2014b). While GM herbicide-tolerant crops have provided easier weed control, GM insect-resistant crops have improved yield in maize and cotton in most countries, except for GM Bt cotton in Australia where the levels of Heliothis sp (boll and bud worm pests) control previously obtained with intensive insecticide use were very good. However, the benefit in Australia has been cost saving and improved environmental impacts due to reduced pesticide use (Brookes and Barfoot 2017).

A meta-analysis of 32 publications (n = 276) has shown that GM maize cultivation led to a significant increase in yield of about 10% as an average, corresponding to 0.7 t/ha, compared to the non-GM isolines or near isolines (Pellegrino et al. 2018). This confirmed an earlier analysis which has shown increases of 0.6 t/ha (Areal et al. 2013), 0.7 t/ha in Spain, 1.1 t/ha in Germany and 1.8 t/ha in South Africa (Finger et al. 2011) for GM over non-GM maize. The biggest yield gain in the Pellegrino et al. (2018) meta-analysis was for quadruple staked hybrids with a 24.5% increase. These yield differences were in part related to the improved percentage change of damaged ears of 73%, 31%, and 84%, for double, triple and quadruple stacked hybrids, respectively. However, where the effect of European maize borer (Ostrinia nubilalis) is not severe maize producers may not benefit from the use of these Bt hybrids (Ma and Subedi 2005). Similarly, a comparison of glyphosate-tolerant soybean cultivars with non-glyphosate-tolerant cultivars indicated a 5–10% yield depression indicating that cultivar choice should be based on balancing the expected weed pressure, relative to the cost of using herbicides and seed of glyphosate-tolerant cultivars (Elmore 2001; Elmore et al. 2001). However, much of this yield depression occurred in the early development of Roundup Ready cultivars awaiting plant breeding input to ensure these gene constructs were introgressed into elite modern cultivars. In the Central Corn Belt of the USA GM maize cultivars have realised increased yields, but for soybean GM cultivars have slightly reduced yields (Xu et al. 2013). In that study, Xu et al. (2013) predicted that the combined effects of these yield trends and GM adoption are predicted to fall short of the growth rate envisioned by industry projections. However, 10 years later there is no published evidence of reduced use of GM crops in these Central Corn Belt states. Additionally, there are studies across a range of environments clearly demonstrating that Roundup Ready 2 Yield® soybean (MON 89788) has no adverse effects compared to a conventional control soybean variety, A3244 (Horak et al. 2015). It largely accepted that modern herbicide-tolerant GM crop cultivars are similar to their non-GM counterparts for yield.

Benefits from GM crops, especially in terms of increased yields, are greatest for the mostly small farmers in developing countries, who have benefitted from the spill over of technologies originally targeted at farmers in industrialised countries (Huang, Rozelle, et al. 2002; Bennett et al. 2006; Carpenter 2010). The average yield increases for developing countries range from 16% for insect-resistant maize to 30% for insect-resistant cotton, with an 85% yield increase observed in a single study on herbicide-tolerant corn.

Crop management

The use of herbicide-tolerant GM crops has been to the primary benefit of farmers providing an option for improved ease of management (Marra et al. 2004; Bonny 2011; Green and Owen 2011). It has been argued that with 95% of the USA soybean market dominated by GM cultivars this reflects the ease of use to the US farmer. For example, growing herbicide-tolerant GM soybean uses a no-till management system leading to reduced energy use and a significant reduction in the use of active chemical ingredients on the soybean crop (down by 27,000 tonnes per year) (Livermore and Turner 2009). In the USA Corn Belt it has been documented that GM technology has a stronger impact on the lower yielding maize crops/areas by reducing exposure to potentially negative outcomes (Chavas et al. 2014). Additionally, and somewhat surprisingly it was found that while the negative ‘maize after maize’ effect still exists for the conventional hybrids, it is reduced to negligible levels for the GM technology in the lower end of the yield distribution. The increase in herbicide-resistant weeds has led to a call for alternative strategies that should not rely solely on different herbicide mechanisms of action, although acknowledging the substantial environmental benefits of herbicide-tolerant GM crops (Reddy and Nandula 2012).

Crop product composition and quality

For many years after the first commercial GM crops were grown and harvested the compositional equivalency between GM crops and non-GM comparators has been a fundamental component of human health safety assessment. Several reviews have been undertaken of studies targeted towards determining the impacts of GM technologies on crop composition and quality. A substantial literature review in 2013 concluded that unintended compositional effects that could be caused by genetic modification have not materialised (Herman and Price 2013). A meta-analysis of 32 publications, over 21 years of field trials, has shown that GM maize cultivation led to a significant increase in yield but no difference in concentration of proteins, lipids, acid detergent fibre, neutral detergent fibre and total dietary fibre in grain compared with isolines or near isolines (Pellegrino et al. 2018). Similarly, a total of 241 statistical comparisons between the multi-trait GM maize crops and their corresponding non-GM controls demonstrated compositional equivalence for both forage and grain (Ridley et al. 2011). And yet another review has questioned these equivalency conclusions for analytes between GM and non-GM cultivars (Herman et al. 2019, 2020), and others have proposed new tests with improved statistical power (van der Voet and Paoletti 2019; Engel and van der Voet 2021). They however do concede that in a practical case study, results were broadly similar, and the improvements were mainly relevant for analytes with hardly any variation between the reference genotypes.

In maize, comparison of seven GM stacks containing event DAS-Ø15Ø7-1 (which expresses Cry1F insecticidal protein and the phosphinothricin N-acetyltransferase (PAT) herbicide-tolerant enzyme) and their matched non-GM near isogenic hybrids showed non-GM breeding more strongly influenced crop composition than did transgenesis or stacking of GM events (Herman et al. 2017). Similarly, maize hybrids with stacked herbicide tolerance and insect resistance (MON 87427 × MON 89034 × NK603) showed compositional equivalence to a near-isogenic non-GM comparator (Venkatesh et al. 2014). The same has been shown not just for maize but also for soybean indicating compositional equivalence for the stacked trait products, reflecting the conclusions obtained for the single trait products (Bell et al. 2018). For soybean both glyphosate and dicamba resistant GM cultivars have been shown to be compositionally equivalent to non-GM soybean (Berman et al. 2011; Taylor et al. 2017). While lignin concentration in leaves and stems did not differ between GM hybrids and their isolines, biomass loss from crop residues was significantly increased in GM hybrids. Conversely, transgene stacking in GM maize has shown 22 proteins mostly associated with energy/carbohydrate metabolism that were statistically differentially modulated (Agapito-Tenfen et al. 2014). The biological relevance and implications of such changes was not described.

One significant unintended benefit of Bt GM maize (single stacked Cry1Ab hybrids) is the lower concentrations of grain mycotoxins such as fumonisins and thricotecens compared with non-GM maize (Ostry et al. 2010). This has been valued as worth about US$23 million annually in the USA (Wu 2006).

Agronomic and phenotypic plant traits

A replicated field trial, undertaken by Bayer Crop Science in Brazil, comparing single or stacked GM materials and the respective non-GM counterparts used as controls at six sites, has shown that combining GM events through conventional breeding does not alter agronomic or phenotypic plant characteristics of soybean (7 events), maize (6 events) and cotton (6 events) crops with stacked genes compared to non-GM equivalents (Jose et al. 2020). Similarly, field trials by Dow AgroSciences also showed no agronomic differences between maize hybrids developed through stacking of 4 individual transgenic events containing the cry1A.105 and cry2Ab2 (MON 89034), cry1F and pat (TC1507), cp4 epsps (5-enolpyruvylshikimate-3-phosphate synthase) and aad-1 transgenes and non-GM near-isogenic hybrids (de Cerqueira et al. 2017). This would indicate that risk assessment of the individual events is sufficient to demonstrate the safety of the stacked products (Clawson et al. 2019). However, it should be acknowledged that it depends on what is changed in the plant by the GM event. However, even when agronomic differences have been documented between GM maize hybrids and their non-GM equivalents it has been concluded that none of the phenotypic differences were expected to contribute to a biological or ecological change that would result in an increased pest potential or ecological risk when cultivating these GM hybrids (Heredia Dıaz et al. 2017).

Improved resistance to pests

Plant genetic resistance to major and minor pests of most crops has been a key goal for plant breeders. However, there are limits within genomes to how successful that strategy can continue to be with changing pest populations and increased agricultural intensification (driven by economic and environmental (slowing the encroachment into natural ecosystems) needs). The advent in 1996 of Bt GM plants expressing pesticidal proteins has provided an opportunity to enhance plant resistance as part of integrated pest management strategies (Kennedy 2008; Kos et al. 2009 Frisvold and Reeves 2010; Hillocks 2014; Machado et al. 2020). A principal advantage of Bt insecticides is that they are generally not harmful to humans, non-target wildlife, or beneficial arthropods (Ortman et al. 2001; Kadoić Balaško et al. 2020). Seeking a wider pool of insect control proteins, other than Bt proteins, has been promoted as likely to improve durability, spectrum of activity, reliability, environmental impact and public acceptability of GM crops (Hilder 2003). But to date this does not seem to have occurred (Table 2).

Environmental benefits

The application of well-managed GM crops has led to improved environmental sustainability (Sharma et al. 2022) due to a reduction in the use of synthetic chemicals, less soil disturbance, and with higher productivity a reduced need for more land to be claimed for agricultural production. Others have concluded that there is no ecological risk in growing GM crops (Raven 2010). Using controlled environments to assess benefits and risks can be misleading and the use of a more holistic approach where assessment is made on the performance of GM plants in the relevant field environment (Liu and Stewart 2019).

Reduced use of synthetic chemicals

While synthetic pesticides have been required to ensure economic crops yields (Pimentel 2005), they have also resulted in ecological consequences of concern potentially affecting non-target species, reducing, and possibly contaminating food sources of other organisms (Devine and Furlong 2007), and waterways (Rosic et al. 2020; Rasool et al. 2022). Additionally, it has been observed that Integrated Pest Management concepts have not necessarily reduced overall pesticide use (Devine and Furlong 2007; Mabubu et al. 2016). However, there is consistent evidence of reduced pesticide use associated with the growing Bt GM crops (Huang, Hu, et al. 2002; Huang et al. 2003; Bennett et al. 2004a; Huang et al. 2005; Qaim and De Janvry 2005; Bennett et al. 2006; Cattaneo et al. 2006; Carpenter 2010; Huang et al. 2010; Lu et al. 2012; Gruissem 2015; Nalluri and Karri 2020), including when genes for herbicide tolerance and pest resistance are stacked (Wossink and Denaux 2006). A meta-analysis by Klümper and Qaim (2014) demonstrated that on average, GM crop adoption has reduced chemical pesticide use by 37%, increased crop yields by 22%, and increased farmer profits by 68%. Yield gains and pesticide reductions are larger for insect-resistant crops than for herbicide-tolerant crops. Yield and profit gains are higher in developing countries than in developed countries.

Environmental Impact Quotient (EIQ) used to assess the broader impact on the environment (plus impact on animal and human health) distils the various environmental and health impacts of individual pesticides in different GM and conventional production systems into a single ‘field value per hectare’ and draws on key toxicity and environmental exposure data related to individual products (Kovach et al. 1992). Negative environmental impacts have been significantly reduced through using herbicide tolerant and pest resistant GM crops achieving on average across 7 GM crops a 23% reduction in environmental impact, in part resulting from 620 million kgs of active ingredient not being applied (Brookes and Barfoot 2017) (Table 4). Even with the slightly increased use of herbicides on GM herbicide-tolerant soybeans the EIQ reduction was still significant due to the switch to more environmentally benign herbicides. The use of pesticide resistance GM crops has been estimated to have saved 671 million kg. active ingredient (a.i.) of pesticides globally between 1996 and 2016, an 8.2% reduction in pesticide use (ISAAA 2018b). Comparing canola production in Western Canada from 1995 and 2006, the environmental impact of herbicides applied to canola was estimated to have decreased by 53%, producer exposure to chemicals by 56%, and quantity of active ingredient applied by 1.3 million kg. due to the widespread use of GM herbicide-resistant canola (Smyth et al. 2011).

Table 4. Impact of changes in the use of herbicides and insecticides and subsequently calculated change in Environmental Impact Quotient (EIQ) from growing GM crops globally 1996–2015 (Slightly modified from Brookes and Barfoot 2017).

GM trait	Absolute change		Percentage change		Area planted (2015) (million ha)
	Weight active ingredient (million kg)	Field EIQ (million EIQ/ha units)	Active ingredient use	Environmental impact
Herbicide-tolerant soybean	+15	−8112	+0.5	−14	76.7
Herbicide and insect-resistant soybean	−3	−348	−4.3	−4	22.3
Herbicide-tolerant maize	−226	−7315	−8.4	−12	45.3
Herbicide-tolerant canola	−25	−854	−18.1	−30	8.6
Herbicide-tolerant cotton	−25	−629	−7.6	−10	4.2
Insect-resistant maize	−87	−3891	−53.3	−58	46.1
Insect-resistant cotton	−268	−11,949	−29.1	−31	21.5

A review 10 years after GM crops were imported and used in Western and Central Europe (i.e. maize, oilseed rape, and soybean) with the two GM traits that are herbicide tolerance and insect resistance there was no scientific evidence (from peer-reviewed scientific journals, scientific books, reports from regions with extensive GM crop cultivation, as well as reports from international governmental organisations) that the cultivation of commercialised GM crops has caused environmental harm (Sanvido et al. 2007).

An alternative view based on data obtained in the USA over the first 4 years of use of Bt and herbicide-tolerant GM crops is that it depends on how you measure synthetic pesticide/ herbicide use (Benbrook 2001). This study indicates that herbicide-tolerant GM cultivars have modestly increased herbicide use and that Bt GM cultivars perpetuate a heavy reliance on insecticide treatments. This partly agrees with the data in Table 4 which indicates a small increase in use of herbicide on herbicide-tolerant GM soybean. However, rather than condemning the technology the advice from Benbrook (2001) was for ‘biotechnology to lead the way toward prevention-based biointensive pest management systems that rest largely on low-impact ways to manage natural biocontrol processes and interactions’. The emphasis here is for long-term pest management benefits from agricultural GM technologies to be through process – and management-based activities, as opposed to those using a product-based system (Benbrook 2000).

Lower C footprint and reduced greenhouse gas emissions

The use of GM crops has been considered to reduce the C footprint of cropping through either reduced use of petrochemicals resulting from fewer herbicide and pesticide applications and reduced need from soil cultivation and/ or the use of no-till or reduced-till allowing carbon to remain in the soil (Brookes and Barfoot 2017; Aslam and Gul 2020; Brookes and Barfoot 2020c). Brookes and Barfoot (2020c) estimated that over that time 34 million tons of CO₂ was not released from reduced fuel use associated with growing GM crops. Using the difference between observed increased yields per hectare due to using GM crops and the counterfactual hectarage needed to generate the same output without the yield boost from GM it was determined that GM crops saved 13 million hectares of land from conversion to agriculture in 2010 alone, and averted emissions that were equivalent to roughly one-eighth of the annual emissions from automobiles in the US (Barrows et al. 2014b).

Reduced tillage

Reduced tillage has been considered to have a positive effect on the soil microbiome and soil structure (Frisvold et al. 2009; Fernandez-Cornejo et al. 2012), but that is not universally accepted (Janušauskaite et al. 2013; Schlüter et al. 2018). Despite these divergent views one benefit that has been heralded for using GM crops is the reduction in tillage used in non-GM crops for managing weed incursion (Marra et al. 2004), resulting in maintaining levels of sequestered soil C (Lee et al. 2014; Hussain et al. 2021) particularly in the surface layers (Deen and Kataki 2003; Brown et al. 2021). It has been estimated that about 302 million tons of CO₂ was not released from soils from 1996 to 2018 (i.e. averaging 14 million tons CO₂ per year) due to using GM crops but they caution that this number was reached with several assumptions being made due to lack of data (Brookes and Barfoot 2020c). Additionally, this is very small proportion of the CO₂ released to the atmosphere each year from all sources, estimated to be 42 × 10⁶ million tons of CO₂ per year (Friedlingstein et al. 2020).

In a review examining the changes in herbicide use in relation to GM canola production in Western Canada, comparing 1995 and 2006 it was noted that with the use of GM canola production 64% of producers are now using zero or minimum tillage as their preferred form of crop and soil management (Smyth et al. 2011). In Mississippi and North Carolina, USA 33% of growers shifted to more conservative tillage practices after the adoption of a glyphosate-tolerant GM crop; this was higher (45%) for those growing cotton than soybean (23%) (Givens et al. 2009). Others monitoring tillage practices in the USA have indicated that more than 65% of glyphosate-tolerant maize, 70% of glyphosate-tolerant cotton, and 50% of glyphosate-tolerant soybean areas received some tillage (Dill et al. 2008). Tillage of course is used for multiple purposes ranging from seed-bed preparation to weed management.

Societal effects in developing economies

Accusations of farmer suicides associated with growing Bt cotton in India (Ho 2010) have proven to be incorrect (Gruère and Sengupta 2011). Another study has shown that using Bt cotton in fact increases returns to labour, especially for hired female workers, increases incomes, including for poor and vulnerable farmers and so concluded that Bt cotton contributes to poverty reduction and rural development (Qaim and Zilberman 2003; Subramanian and Qaim 2010). Comparison of Bt cotton grown either as official or unofficial hybrids (i.e. second generation F2 seed) showed increased yield for the official hybrids and reduced insecticide for both official and unofficial hybrids compared with non-GM equivalents (Bennett et al. 2005). Data collected in India between 2002 and 2008 and analysed to control non-random selection bias in technology adoption has shown that Bt GM cotton has caused a 24% increase in yield per acre through reduced pest damage and a 50% gain in cotton profit among smallholders (Kathage and Qaim 2012). However, other commentaries suggest that Bt technology has failed in reducing the use of insecticides in India largely due to the emergence of secondary pests (Viswanathan and Lalitha 2010; Kranthi and Stone 2020).

Elsewhere, in the Philippines the use of Bt maize has been shown to increase net farm income, off-farm income, and household income and reduces the probability of a household's income falling below the poverty line (Yorobe and Smale 2012). In China, farmers using Bt cotton reported reduced use of pesticide without reducing yield per hectare or quality of cotton harvested (Pray et al. 2001; Huang, Hu et al. 2002; Huang et al. 2005). Farmers also reported fewer pesticide poisonings than those using non-GM cotton. A similar outcome was found in India where the use of Bt cotton helps to avoid several million cases of pesticide poisoning which also results in sizeable health cost savings (Kouser and Qaim 2011). While GM crops are not going to solve all problems in developing countries, they do hold significant potential to contribute to poverty reduction, better nutrition and health, and sustainable development (Qaim 2010; Tonukari and Omotor 2010). In Africa, the advent of GM crops has been viewed as a potential benefit to subsistence farmers (Bennett et al. 2004a; Bennett et al. 2006; Muzhinji and Ntuli 2020; Gbashi et al. 2021), largely because the technology is delivered in the seed (Ehirium et al. 2020). However, there is an alternative opinion that these farmers should be supported to use their local natural ecosystems or agro-ecological approaches that have been successfully used to produce food in the past (Kisiangani and Pasteur 2008). Indeed the introduction of GM technologies is likely to be more successful if using indigenous innovation rather than simply ‘transporting’ agricultural technology from developed to developing countries (Shen 2010).

GM product quality and safety testing for consumption by either animals or humans

Providing confidence on the safety of derived food for human consumption and feed for animals is paramount for any technology to be trusted and accepted. Indeed the consumption of any modified or new foods either by animals or humans need to be effectively and thoroughly tested for their impact on health and welfare. National risk-based food control systems are essential to protect the health and safety of the public (Gizaw 2019). Concerns about the safety of food and feed from GM crops are often exacerbated by opinion rather than fact (Burke 2004) have led to debates resulting in polarised views which result in confusion and inevitably mistrust. This topic of food safety is rightly one that should be closely examined when considering the use of GM crops. Regulatory approvals require testing for safety when consumed as food (van den Belt 2003; EFSA GMO Panel 2008; Hull et al. 2010; Baulcombe et al. 2014). One would assume that no approval for use would be given if this testing determined some concerns. Yet in some studies undertaken after these crops were commercialised concerns have been raised (Table 5). So who is to be believed?

Table 5. Summary of toxicology and health testing trials involving comparisons of GM crops and their non-GM equivalent indicating (A) some concerns and (B) little concern. Also refer to Magana-Gomez and Calderon de la Barca (2009) and DeFrancesco (2013) for additional information.

GM crop	Code or name and transgene	Type of study	Result compared to non-GM equivalent	Reference
A. Studies indicating some concern from toxicology and health testing trials
Maize	MON863 – Bt expressing Cry3Bb1	90 d rat feeding trial	Slight weight decrease and hepatorenal toxicity	Séralini et al. (2007)
	MON810 – Bt expressing Cry1Ab	30 and 90 d mouse feeding trial	Induced alterations in intestinal and peripheral immune response of weaning and old mice	Finamore et al. (2008)
	Bt	3 generations of feeding to rats	Some minimal histopathological changes in liver and kidney; no difference in organ weights	Kilic and Akay (2008)
	MON810 and MON863 – Bt expressing Cry1Ab and Cry3Bb1, respectively	Up to 90 d rat feeding trial	Negative impact on the function on kidneys and liver	de Vendomois et al. (2009)
	NK603 – glyphosate and AMPA tolerant	Up to 90 d rat feeding trial	Negative impact on the function on kidneys and liver	de Vendomois et al. (2009)
	NK603 – glyphosate tolerant	2-year feeding trial with rats	Females developed mammary tumours before controls; Males had four times more large palpable tumours starting earlier than in the control group	Séralini et al. (2014)
Peas (Pisum sativum)	Expression of Bean α-Amylase Inhibitor	Mice	Elicited specific immunoreactivity of these proteins	Prescott et al. (2005)
Potato	Expressing CryI gene	Mice	Mild changes are reported in the structural configuration of the ileum of mice	Fares and El-Sayed (1998)
	Expressing the lectin Galanthus nivalis agglutinin (GNA)	Rats	GNA expression was associated with significantly greater mucosal thickness of the stomach when compared with non-GM potato diets	Ewen and Pusztai (1999)
Soybean	Insertion of the bacterial CP4 EPSPS (5-enolpyruvylshikimate-3-phosphate synthase) gene conferring a high level of tolerance to glyphosate	Mice	GM soybean intake can nuclear features; however mechanisms responsible for this remain unclear	Malatesta et al. (2002)
	Insertion of the bacterial CP4 EPSPS (5-enolpyruvylshikimate-3-phosphate synthase) gene conferring a high level of tolerance to glyphosate	Mice	GM soybean can influence the pancreatic metabolism in mouse. Unfortunately, the reasons for such modifications remain so far unclear.	Malatesta et al. (2003)
	Glyphosate tolerance	Mice	In testes of GM-fed mice the number of perichromatin granules is higher and the nuclear pore density lower; could be due to presence of residual glyphosate	Vecchio et al. (2004)
	Not specified	Allergenic skin tests on humans	Skin test results of 49 patients showed 13 positive results to wild soybeans and 8 positive results to GMO soybeans	Yum et al. (2005)
	Insertion of the bacterial CP4 EPSPS (5-enolpyruvylshikimate-3-phosphate synthase) gene conferring a high level of tolerance to glyphosate	Mice	GM soybean intake can influence some liver features during ageing	Malatesta et al. (2008)
	Type GM soybean product, which was from Kyrgyzstan, was not described but PCR revealed the presence of the 35S + FMV promoter	30 d feeding trial with rats	Visible changes in the spleen characterised by an increase in size, average weight, and changes in its structure and colour	Egamberganovich (2020)
B. Studies indicating little concern from toxicology and health testing trials
Maize	MON863 – Bt expressing Cry3Bb1	90 d rat feeding trial	No adverse effects on health, body weight gain, food consumption, clinical pathology parameters (haematology, blood chemistry, urinalysis), organ weights, gross and microscopic appearance of tissues	Hammond et al. (2006)
	Maize line 1507, containing event DAS-Ø15Ø7-1 (1507), expressing Cry1F	Sub-chronic rat feeding trial over 90 days	No toxicologically significant differences for nutritional performance variables, clinical and neurobehavioral signs, ophthalmology, clinical pathology (haematology, clinical chemistry, coagulation, and urinalysis), organ weights, and gross and microscopic pathology	MacKenzie et al. (2007)
	Bt176 expressing Cry1Ab	Sheep feeding trial over 3 years	No adverse effects on their health or performance and that no horizontal gene transfer to ruminal microorganisms or animal tissues was detected	Trabalza-Marinucci et al. (2008)
	Stacked GM crop NK603 x MON810	Mice	Multi-generation study showed no statistically significant differences concerning parental body mass; however time-related negative reproductive effects of the GM maize; some effect on some metabolic pathways	Velimirov et al. (2008)
Mix of maize and soybean	GM maize – a combination of NK603, MON863 and MON810 genes (expressing the CP4 EPSPS, Cry 3Bb1 and Cry 1Ab proteins respectively), GM soybean was 100% Roundup Ready (expressing the CP4 EPSPS protein),	Long-term toxicology study of 22.7 weeks with isowean pigs	No differences between pigs fed the GM and non-GM diets for feed intake, weight gain, mortality and routine blood biochemistry measurements, but there was a slight increase in severe stomach inflammation amongst male and female pigs	Carman et al. (2013)
Potato	Cry1Ab gene	90 d rat feeding trial	No effect on growth rate, food intake, food efficiency, and general health status; no effect on serum concentrations of immunoglobulin A (IgA), IgG, IgM, orinterleukin 6	Rahnama et al. (2017)
Rice	Cry1Ab/Ac gene	50 and 110 d feed trial with rats	Exogenous gene and protein of the transgenic rice did not remain in the intestines	Xu et al. (2015b)
Soybean	Glyphosate tolerance	105 d feeding trials with mice and rats	No effect on growth, feeding value and histopathology of immune-related organs	Teshima et al. (2000)
	Glyphosate tolerance	90 d feeding trial with mice	No effect on macromolecular synthesis or cell growth and differentiation of testicular cell populations; no differences in litter sizes and body weights	Brake and Evenson (2004)
	Glyphosate tolerance	90 d feeding trial with rats	No differences for gross necropsy findings, haematological or urinalysis values and clinical serum parameters	Zhu et al. (2004)
	Not specified but did include 35S promoter and used soybean meal (solvent extracted)	Rabbits	35S CMV was not found in all samples of serum, heart, skeletal muscle, liver and kidney; no differences detected for body weight, fresh organ weights	Tudisco et al. (2006)
	Type of GM soybean not specified in abstract (article in Russian)	Comparison of third-generation female rats where both they and their mothers	No definite negative or positive impact of GM soy components on their physiological state compared with animals fed non-GM soybeans	Dolaychuk et al. (2013)
	MON87708 – dicamba tolerant	90-day sub-chronic toxicology study with rats	No indication of adverse effects associated with the consumption of dicamba-tolerant soybean	Wang et al. (2016)

It should be noted that many non-GM crops also contain potentially toxic or allergenic compounds (Halford and Shewry 2000; Purchase 2005). These may result either from the plant itself or from fungal activity before and after harvest leading to mycotoxin production (Miller 2008; Chulze 2010). It has been argued that food from GM crops may be safer than food derived from non-GM crops, largely because the risks associated with GM crops are readily quantified and monitored as part of the rigorous assessment system that goes beyond that applied to non-GM derived foods (Halford and Shewry 2000).

GM feed for animals

Most output from GM crops is used in animal feed rather than human food. An estimated 70%–90% of all GM crops, principally soybean and maize, but also including cotton and canola, are used to feed animals (Flachowsky et al. 2012; Ritchie and Roser 2021) with the biggest users being USA, China, and Europe (Baulcombe et al. 2014). For GM soybean, the biggest producers are USA and Brazil, followed by Argentina (Qaim and Traxler 2005) with most soybean used in China (Hong et al. 2013) and Europe being imported (Lucht 2015; Ritchie and Roser 2021; Shahbandeh 2022). Animal feed of soybean (77% of global use) is directly from soybean grain/seed rather than as soybean cake or meal which is the by-product after processing for oil production (Ritchie and Roser 2021). While maize is the staple diet for many the steady increase of maize production over the years has corresponded to an increase in the use of maize for animal feed (Ranum et al. 2014) now making up 56% of total use globally, with food use only at 13% and processing, post-harvest losses and other uses at 30% (Erenstein et al. 2022).

For testing the safety of GM crops as feed for animals, tests can be undertaken with the actual animal and not laboratory animals (rats (Rattus spp.) and mice (Mus musculus)). The general conclusion is that most GM crops used for animal feed have input traits that do not change their composition or nutritional value for animals and feeding GM crops does not result in detection of transgenic DNA or their translated proteins in meat, milk, or eggs (Flachowsky and Chesson 2003; Van Eenennaam 2013; Van Eenennaam and Young 2014; Deb et al. 2016; Vicini 2017; de Vos and Swanenburg 2018; Akram et al. 2019; Blair and Regenstein 2020).

The widespread use of glyphosate-resistant crops contributed to weeds being resistant to glyphosate-based herbicides which in turn has led to increased levels of herbicide applications (Jarrell et al. 2020). Higher levels of herbicide use may have increased glyphosate residue concentrations in animal feed. So the issue here is the accepted toxicity of glyphosate. Some commentators believe that current safety standards for glyphosate-based herbicides are outdated and may fail to protect public health or the environment (Cuhra 2015; Vandenberg et al. 2017 Novotny 2022), yet others believe there is ‘very limited evidence for an association between glyphosate-based formulations and non-Hodgkin lymphoma, and an overall inconclusive for a causal or clear associative relationship between glyphosate and cancer in human studies’ (Portier et al. 2016). A complete review of the controversy about the safety profile of the herbicide glyphosate and its commercial formulations is beyond this current publication. However, in a 2017 review concluded that often the expressed concern about health risks associated with long-term exposure to glyphosate were not supported by the available scientific evidence (Mesnage and Antoniou 2017). They, however, concluded that while evidence exists that glyphosate-based herbicides are toxic below regulatory set safety limits, some arguments serve to distract rather than to give a rational direction to much needed future research investigating the toxicity of these pesticides, especially at levels of ingestion that are typical for human populations. Other reviews have concluded that epidemiological studies report contradictory data, but findings indicate that exposure to glyphosate is not completely safe (Davoren and Schiestl 2018; Agostini et al. 2020; Soares et al. 2021).

Early studies concluded that glyphosate-tolerant GM soybean had similar feeding value as non-GM equivalents when fed to chickens (Gallus gallus), catfish (Siluriformes spp.), and dairy cattle (Bos taurus) (Hammond et al. 1996). Similarly, glyphosate-tolerant maize has been shown to be equivalent in growth, feed efficiency, adjusted feed efficiency, and fat pad weights when fed to poultry compared to non-GM maize (Sidhu et al. 2000). A review of impacts on glyphosate in animal feed has concluded that glyphosate use in GM crops fed to poultry and livestock has neither affected animal health, rumen/gut microbes or production nor the safety of consuming meat, milk, and eggs (Vicini et al. 2019). However, while glyphosate has been considered animal safe a recent study reviewing controlled laboratory trials has concluded that glyphosate-based herbicides may cause adverse effects in animal reproduction, including disruption of key regulatory enzymes in androgen synthesis, alteration of serum levels of oestrogen and testosterone, damage to reproductive tissues and impairment of gametogenesis (Jarrell et al. 2020). They concede that there have been very few published studies investigating the effects of exposure to glyphosate-based herbicides on agriculturally important animals, and the majority of those that have, have given no concern to reproductive health and performance. However, assuming there is a potential issue here they propose that including 0.2% humic acids in the feed would neutralise the potential impact of glyphosate on reproductive fitness in livestock. A solution demonstrated effectively by Shehata et al. (2014) with broiler chickens. A critical review on the behaviour, fate, and detrimental impacts on ecological and human health of glyphosate concluded that glyphosate can be judiciously used in agriculture with the inclusion of safer surfactants in commercial formulations than the surfactant, polyethoxylated tallow amine, which is toxic itself and is likely to increase the toxicity of glyphosate (Meftaul et al. 2020).

While most of the focus has been on the safety of glyphosate-resistant GM crops there have been GM crops developed with resistance to other herbicide chemistries such as glufosinate (or phosphinothricin), 2,4-D, dicamba, isoxaflutole, mesotrione, oxynil and sulfonylurea (Waltz 2015a; Kumar et al. 2020). Maize grain from plants containing event DP202216, which confers herbicide tolerance through expression of phosphinothricin acetyltransferase when fed to rats for 90 days showed no biologically relevant effects or toxicologically significant differences compared with rats fed diets containing control non-GM maize grains (Carlson et al. 2020). Similar results have been obtained for other GM maize events expressing the phosphinothricin-N-acetyltransferase gene (MacKenzie et al. 2007). Canola event MON94100 with tolerance to dicamba was determined to be safe for consumption as food and feed compared with the non-GM cultivars (EFSA GMO Panel et al. 2022b).

A review undertaken in 2013 of all available GM crop whole food studies indicated no convincing toxicological concern and called into question the adequacy, sufficiency, and reliability of safety assessments based on crop molecular characterisation, transgene source, agronomic characteristics, and/or compositional analysis of the GM crop compared with near-isogenic lines (Bartholomaeus et al. 2013). They concluded that because safety predictions based on crop genetics on compositional analyses were in complete agreement with well-conducted animal testing that whole food animal toxicity studies are unnecessary and scientifically unjustifiable. Similarly, pigs (Sus scrofa domesticus) fed GM soybean meal (Roundup Ready MON 40-3-2) and Bt maize (MON 810) showed no effect of GM components on body weight gain, feed utilisation, and carcass and meat quality compared with non-GM feeds (Świątkiewicz et al. 2013).

Feeding Bt pest resistant maize as silage to dairy cows showed no significant treatment effects on milk yield, milk composition, and yield of milk constituents, and the dry matter intake of the GM cultivar was not significantly different from commercial non-GM cultivars (Donkin et al. 2003; Phipps et al. 2005). A trial with Japanese quails (Coturnix japonica) fed GM rice containing Cry1Ab/1Ac gene for two consecutive generations showed no significant effect on growth and development (Li et al. 2015). Similarly, work comparing Bt GM rice fed to chickens showed no adverse effect on aspects of immune function during 42-d feeding trial (Liu et al. 2016). Similarly, Bt maize (MON810) expressing Cry1Ab protein showed compositional and nutritional equivalence to its isogenic non-GM comparator in a long-term feeding study with lactating dairy cows (Steinke et al. 2010). Additionally, no residues of the Cry1Ab/Ac gene were found the intestine of chickens fed transgenic rice for 21 and 42 d, and the Cry1Ab/Ac gene gradually degraded after 4 h in the gut in vitro test (Xu et al. 2015a). Feeding of Cry1Ab protein in Bt MON810 maize to sows during gestation and lactation and their offspring from weaning to 115 days postweaning showed no detrimental effects on intergenerational pig growth and health (Buzoianu et al. 2013a, 2013b).

A review undertaken after 20 years of using GM crops concluded that GM crops with input traits (involving herbicide tolerant and pest resistant genes) can be considered as substantially equivalent to their isogenic counterparts (Flachowsky 2017). GM crops with output traits which may influence composition and quality of food of animal origin and in the case of lower content of undesirable substances may in fact improve the feed value of GM crops and feed derived from them. This conclusion was built on the foundation of extensive trials across a range of farmed animal species (Flachowsky et al. 2005, 2007, 2012).

Lucerne transformed with the gene encoding for Betaine Aldehyde Dehydrogenase (BADH) which confers tolerance to salinity and drought stresses was compared with non-GM lucerne by supplementing 50% of the diet of rabbits (Oryctolagus cuniculus) (Zhong et al. 2014). No adverse effects were observed in rabbits consuming GM lucerne and non-GM lucerne with the feeding value of GM and non-GM lucerne being equal.

Food safety for human consumption

The advent of GM crops has led to concerns about food safety and the need for rigorous testing (Nordic Working Group 1991; OECD 1993, 1998; Pusztai 2001, 2002). To manage potential risks from using GM crops there has nearly always been a framework of science-based risk assessment and risk management measures in place to oversee their commercialisation (Craig et al. 2008). However, mistrust of companies and corporations developing GM crops who claim they have scrutinised and evaluated new cultivars to show that they are as safe as earlier crops produced have made these claims unacceptable (McHughen 2000). Independent reviews of risk evaluations are required. In 2007, it was considered that there were relatively few published accounts showing that GM crops and food are toxicologically safe (Domingo 2007). However, a decade later it was concluded after statistical reanalysis and review of experimental data presented in some studies claiming possible GM technology-related health concerns that quite often, in contradiction with those authors’ conclusions, the data actually provides weak evidence of harm that cannot be differentiated from chance (Panchin and Tuzhikov 2017). They surmised that statistically unaccounted multiple comparisons have led to some of the most cited anti-GM organism health claims in history. Further to that a review of scientific literature on GM crop safety in the first 10 years from commercialisation in 1996, concluded that the scientific research conducted so far had not convincingly detected any significant hazards directly connected with the use of GM crops (Nicolia et al. 2014).

It was documented early in the development of GM crops that just showing that GM food is chemically similar to its non-GM counterpart is not adequate evidence that it is safe for human consumption (Millstone et al. 1999). Food safety testing often involves the use of feeding trials using rodents. It has been claimed that safety assessment of foods derived from GM crops has been undertaken with due care, more care than occurs from food derived from conventionally bred plants (Kok et al. 2008). Extensive testing prior to release ensures that food from GM crops is at least as safe as new non-GM cultivars (Lack 2002; Konig et al. 2004), which however are not required to be tested for being safe to eat.

The generally accepted conclusion is that there has been no evidence of ill effects linked to the consumption of any approved GM crop (The Royal Society 2016; Ladics 2019). Acute feeding studies involving Cry protein (used in Bt GM crops) consumption have shown no adverse effects, they are non-toxic to humans and pose no significant concern for allergenicity (Betz et al. 2000; Carzoli et al. 2018). However, there have been several trials showing some adverse effects summarised in Table 5. Equally there have been many studies showing no meaningful or significant effects of GM crops in similar types of rodent feeding trials (Table 5). Interestingly, in a review of 30 published studies between 2000 and 2010 it was also found that the numbers of studies showing no harm are now roughly equal to those showing harm when comparing GM and non-GM crops in toxicology trials (Domingo and Bordonaba 2011). However, they did make the point that most of the studies demonstrating that GM foods are as nutritional and safe as those obtained from non-GM crops, have been undertaken by biotechnology companies or associates. A review undertaken 20 years after the first release of GM crops also concluded that there are no case-reports of allergic reactions or immuno-toxic effects resulting from GM feed consumption as compared with non-GM feed (De Santis et al. 2018). Yet another review (Shen et al. 2022) of one crossover trial in humans and 203 animal studies from 179 articles used a subjective classification system to identify 21 adverse events involving 7 GM events all of which have had regulatory approval for use in some countries/regions.

Some of the studies summarised in Table 5 which indicated potential issues related to health and safety with feeding GM crops require more detailed analysis and commentary. Séralini et al. (2007) concluded that ‘longer experiments are essential in order to indicate the real nature and extent of the possible pathology; with the present data it cannot be concluded that GM maize MON863 is a safe product’. However, this group in a further publication (de Vendomois et al. 2009) commented that ‘these signs of toxicity alone do not constitute proof of adverse health effects’. Kilic and Akay (2008) concluded that ‘long-term consumption of transgenic Bt maize throughout three generation did not cause severe health concerns on rats’. de Vendomois et al. (2009) concluded that ‘our data presented here strongly recommend that additional long-term (up to 2 years) animal feeding studies be performed in at least three species, preferably also multi-generational, to provide true scientifically valid data on the acute and chronic toxic effects of GM crops, feed, and foods’. Finamore et al. (2008) concluded that 'the significance of these data remains to be clarified to establish whether these alterations reflect significant immune dysfunctions, these results suggest the importance of considering the gut and peripheral immune response to the whole GM crop’. Trabalza-Marinucci et al. (2008) concluded that ‘cytochemical modifications of the gastrointestinal organs and the immune response mechanisms that take place in GM-fed animals should deserve special emphasis and priority in future investigations’. Malatesta et al. (2008) concluded that ‘GM soybean intake can influence the liver morpho-functional features during the physiological process of ageing and, although the mechanisms responsible for such alterations are still unknown and some data have been discussed on a speculative basis, there are several findings underlining the importance to further investigate the long-term consequences of a GM-diet and the potential synergistic effects with ageing, xenobiotics and/or stress conditions’. A long-term (2-year) study on the toxicity of glyphosate GM maize was purported to show mammary, hepatic and kidney disturbances leading to tumours and mortality but was later retracted by the journal based in inconclusiveness (Séralini et al. 2012). Several co-authors to this work are also authors of studies showing detrimental effects of GM crops on health (6 of which are mentioned in Table 5). This work had been heavily criticised due to several key experimental design and statistical flaws which many have concluded provided misleading results (Arjo et al. 2013; DeFrancesco 2013). In addition, a meta-analysis of 24 feeding studies has concluded that long-term studies provide little additional value over 90-day feeding trials used for safety assessment (Snell et al. 2012), a position also taken by others who suggest long-term feeding studies are extremely weak, with no power to distinguish between groups, and have been labelled ‘hypothesis-less fishing trips’ (Chassy 2010).

Two non-peer-reviewed articles, not included in Table 5 but frequently quoted, by Ermakova (2006, 2009) of the Institute of Higher Nervous Activity and Neurophysiology of the Russian Academy of Sciences in Moscow indicated that rats fed diets containing glyphosate-tolerant GM soybeans gave birth to pups with low survival rates or stunted growth. This work was widely used to demonstrate the potential toxicity of GM crops. Others considered this work to ‘depart so dramatically from previously reported findings as to be remarkable’ and as a result require scrutiny (Marshall 2007). Criticisms and concerns include:

• The source of the GM soybean and no evidence of nutritional composition of test and control diet;

• The use of multiple animals per cage accessing the food sources ad libitum;

• Number of animals used per trial group was low at 5, whereas most credible reproductive toxicology studies use 20–25 animals;

• Protocols used for measures of animal weights did not align with other toxicology study methods’

• Behavioural studies do not appear to be performed in a double-blind manner;

• No objective behavioural or morphological data are presented. Claims should not be made without presentation of evidence;

• The timing and causes of death of animals is not reported;

• The high death rate supposedly caused by GM soybean defies credibility and if correct would have been seen in previous studies – high death rates likely to be a result of poor experimental design; and

• A total failure of adult animals fed GM soybean to produce offspring was so remarkable as to be considered unbelievable.

These researchers considered Ermakova’s claims implausible and ‘that no meaningful inferences can be drawn from these results’, and they contradicted peer-reviewed work on the same topic (Teshima et al. 2000; Brake and Evenson 2004; Zhu et al. 2004).

In the studies where disadvantage health effects have been noted from food derived from GM crops (Table 5) the claims are not about the GM method used but rather as a result of the trait-related specific gene introduced (Lack 2002; The Royal Society 2016). When these occur and are identified they would not be moved forward commercially (Lack 2002; Konig et al. 2004). Allergenicity testing (EFSA GMO Panel et al. 2021, 2022a) has resulted in the absence of transgenic proteins in foods that might cause allergic reactions (Lehrer and Bannon 2005). As an example, a gene from Brazil nut (Bertholletia excelsa) when transferred to a soybean cultivar to improve its nutritional value was shown to test positive in the serum assay for allergenicity by cross-reactivity with Brazil nut (Nordlee et al. 1996) and so its development was terminated.

There is no doubt that testing for potential health effects of GM crops is complex and polarising. The situation is well described by DeFrancesco (2013) who concludes that ‘critics and proponents of GM crops alike agree that genetically modified foods have failed to produce any untoward [human] health effects, and that the risk to human health from foods contaminated with pathogens is far greater than from GM crops’. Indeed, extensive reviews of impacts of GM crops such as Bt corn, cotton and maize conclude that they are nontoxic to humans and pose no significant concern for allergenicity (Betz et al. 2000; Sasson 2018). It has also been pointed out that there are many naturally occurring potentially toxic molecules in plants, but virtually none of them is ever present at a concentration that would cause harm (Chassy 2010).

Despite that, opponents of GM crops and food cite potential risks of allergenicity (Bernstein et al. 2003). However, it is standard that during deregulation the allergenicity of the gene product is assessed either against a panel of blood donor samples or via prick tests (Bolhaar et al. 2005; Goodman et al. 2008; Ladics 2008). The lack of allergenicity needs to be demonstrated before a GM crop can be deregulated (Schauzu et al. 2012; Selb et al. 2017; Ladics 2019; EFSA GMO Panel et al. 2017, 2021). In a review of 83 studies (Dunn et al. 2017) none identified that direct consumption of a GM food was associated with an increased rate of clinical allergy, and only three noted increased sensitisation to the GM product, one of which was associated with the Brazil nut allergy and the other two showed no increase in markers in any clinical allergy in mouse models. Currently, there appears to be no serious risk about allergenicity of GM foods produced and consumed, and while current assessment procedures are robust, new information and technologies will continue to refine these procedures (Yavari et al. 2016). Allergenicity testing of GM crops often uses a rodent model but this has been criticised on the basis that not all proteins demonstrate allergic reaction using a digestive model and other tests are required (Verhoeckx et al. 2019).

Food quality and nutritive value

Two issues need to be examined here: (1) the deliberate use of GM technologies to improve the nutritional value of crops, and (2) the impact of transgenes used to provide other traits on crop nutritional quality.

It has been considered that crops could be genetically transformed to improve their nutritive quality and value in an attempt to eradicate or lessen the impact of malnutrition in some countries (Bouis 2007; Farre et al. 2011; Jones 2015; Garg et al. 2018; Medina-Lozano and Díaz 2022). Attitudes towards the use of biofortified foods using supplementation rather than GM technologies has also struggled to be adopted without approved health and nutrition claims (Gannon et al. 2014; Wortmann et al. 2018; Adekambi et al. 2020; Welk et al. 2021). GM approaches have the benefit of being able to enhance the expression of nutritional traits such as mineral availability and vitamin content (Brinch-Pedersen et al. 2007; Zhu et al. 2007) that may not otherwise be available within the plant’s genome. It has been proposed that biotechnology could also be part of the solution to reducing naturally occurring allergies in food rather than a cause of increased concern (Morandini 2010). Garg et al. (2018) summarise numerous examples of crops that have demonstrated improvement in nutritional value through the use of transgenesis. The majority (over 50 events) are still at a research stage but 10 had been released as commercially available technologies (Table 6). It makes good sense that when GM technologies lead to compositional changes that special consideration should be given to their nutritional impacts (De Santis et al. 2018).

Table 6. Summary of commercially released crops with transgenic nutritionally enhanced traits. Modified from Garg et al. (2018) and refer to McGloughlin (2010) and Jones (2015).

Crop	Nutritional trait	Company	Example reference
Canola	Phytate degradation (increase in available P)	BASF	Jones (2015)
	High lauric acid	Monsanto
Cassava (Manihot esculenta)	Iron	Biocassava Plus	Sayre et al. (2011)
	Beta-carotene
	Protein
Maize	Phytate degradation	Origin Agritech, China	Origin Agritech (2010)
	Lysine	Monsanto; Renessen	Frizzi et al. (2008); Tang et al. (2013)
Potato	Reduced amylose and increased amylopectin in starch granules	BASF	Karlsson et al. (2007)
Soybean	Oleic acid	Dupont; Monsanto	Zhang et al. (2014)
	γ-Linolenic Acid + stearidonic acid	Monsanto	Sato et al. (2004)

Comparison of Golden Rice with a non-GM cultivar, Nakding, has shown no significant difference between the two cultivars in population densities of insect pests and natural enemies (Kim et al. 2010). There is a growing public awareness of nutritionally enhanced GM rice (Challa et al. 2020). A lysine enhanced GM maize (LY038) developed to accumulate free Lys in the germ portion of maize grain and also stacked with Bt gene (LY038 X Mon 810) providing protection against European maize borer were fed to broilers over a 42 d period in comparison with non-GM maize. Weight gain, feed efficiency, carcass yield and composition of broilers fed diets containing LY038 or LY038 × MON 810 were not significantly different from that of broilers fed lysine supplemented diets and were significantly superior to that of broilers fed conventional maize diets without supplemental lysine (Lucas et al. 2007). Another transformation event to enhance lysine in maize used the gene for a lysine-rich protein (sb401) obtained from wild potatoes (Solanum berthaultii) to produce Y642 transgenic maize (He et al. 2009). In a rodent feeding trial, Y642 was compared with a near-isogenic non-genetically modified (non-GM) commercially available control quality protein maize (Nongda 108). There were no adverse diet-related differences in body weights, feed consumption/utilisation, clinical chemistry, haematology, absolute and relative organ weights observed. Y642 was judged to be as safe and nutritious as conventional quality protein maize. Processed soybean meal (SBM) from two soybean transgenic cultivars developed for higher protein content were compared with two non-GM cultivars for levels of amino acid, gross energy, lipid, and fibre and digestibility of metabolisable energy and amino acids using cockerels (Edwards et al. 2000). Results indicated that at least one of the GM soybeans had considerable advantages over conventional soybean meal as a feed ingredient for broiler chickens. An evaluation of transgenic lupin seeds (Lupinus angustifolius) with higher contents of methionine using broiler chickens showed that the higher AME value in transgenic lupins may be related to the lower content of soluble non-starch polysaccharides (45.6 vs 60.7 g kg−1 air-dry basis) (Ravindran et al. 2002). The amino acids in transgenic lupins were as digestible as those in non-GM lupins. To improve the level of digestible protein in maize used for animal feed the milk protein porcine α-lactalbumin was transformed into maize and expressed in both callus and kernels (Yang et al. 2002). An upregulation of the carotenoid pathway in Canola using a bacterial phytoene synthase gene (crtB) not only increased carotenoid levels but also fatty acid composition resulting in a higher percentage of oleic acid but decreased percentage of linoleic and linolenic acids (Shewmaker et al. 1999).

In seeking to understand the extent of unintended effects of transgenes on composition and nutritional value the compositional analyses of 129 GM crops have been investigated by the US Food and Drug Administration which showed no significant differences, for any plant compounds believed to have biological relevance when compared to non-GM equivalents (DeFrancesco 2013). Some unintended consequences of GM crops have been described as beneficial. For example, herbicide-tolerant soybeans have been shown to have an overall reduction in phytoestrogen levels of 12%–14% compared to their isogenic comparator crop, mostly attributable to reductions in the concentrations of genistin and, to a lesser extent, in daidzin (Lappé et al. 1999). In GM rice, containing CryIAc and sck (trypsin inhibitor) genes the levels of sucrose, mannitol and glutamic acid were significantly higher than for the non-GM equivalent (Zhou et al. 2009). A comparison of three different GM rice cultivars (one expressing antifungal genes, the second a chitinase gene, and the third Cry1Ac and sck genes) with their non-GM comparators showed some variation in amino acid levels which led to the conclusion that the nutrition value of transgenic rice had decreased (Jiao et al. 2010).

Unintended consequences of traits in GM crops – fact or myth?

While many have condemned GM crops for unintended consequences, some of these claims need scrutiny while others need to be taken seriously and managed appropriately. Indeed if one reflects on non-GM conventional agricultural methods there is no doubt that herbicide and pesticide application can have large and well-documented unintended consequences (Pilson and Prendeville 2004; Rezende-Teixeira et al. 2022). And while ‘two wrongs don’t make a right’ it is worth examining whether the use of GM crops may result in fewer and less concerning, more easily managed unintended consequences. Discussed in the previous section have been the largely unconfirmed unintended consequences of GM crops regarding feed and food safety, and health-related issues. Science commentators soon after the commercialisation of GM crops expressed concern about their impact on the environment (Snow and Palma 1997; Wolfenbarger and Phifer 2000), non-target species (EFSA GMO Panel et al. 2010), and gene flow (Snow 2002). While not dealt with in detail, concerns about glyphosate-tolerant GM crops causing differences in plant mineral nutrition (e.g. Zobiole et al. 2010) and disease susceptibility (Means and Kremer 2007) have been proven incorrect (Duke et al. 2012). Reviews on concerns related to the use of GM crops continued to provide insights and mitigation measures to reduce the ecological impact (Lombardo et al. 2020; Vulchi et al. 2022)

Increased herbicide tolerance of weeds

Herbicide resistance in weeds is a global issue and not just restricted to, or solely caused by, the use of herbicide-tolerant GM crops (Heap 2014). Even in 2014 it was recognised that the lack of new novel herbicide chemistries combined with the rapid increase in multiple resistance in weeds threatens crop production worldwide. This is not simply a herbicide-tolerant GM crop-related issue, although it is acknowledged that the overuse of glyphosate to manage weeds in GM herbicide-tolerant crops has in some situations lead to the development of glyphosate-resistant weeds (Mortensen et al. 2012; Shaner et al. 2012; Brookes and Barfoot 2017). In 2022, there were 56 weeds recognised as exhibiting resistance to glyphosate worldwide, although several are not associated with glyphosate-tolerant GM crops (Weedscience 2022). Where herbicide tolerance of weeds has developed in GM crops (Wakelin and Preston 2006; Ghanizadeh et al. 2019; Rigon et al. 2020) the issue here is more to do with poor management and use of herbicides not the development of GM crops per se. To avoid the development of glyphosate-resistant weeds GM crop growers are being advised to include other herbicides (with different and complementary modes of action, such as dicamba) in combination with glyphosate (Green and Owen 2011; Vencill et al. 2012; Brookes 2014; Brookes and Barfoot 2017; Ravisankar et al. 2017). This would also be a useful strategy for avoiding herbicide resistance in non-GM crops (Beckie et al. 2021).

Overuse of herbicides and pesticides

There have been instances where economically important pests have become resistant to synthetic insecticides. This can lead to overuse of insecticides in a desperate bid to control them (Benbrook 2012). However, this situation can also be resolved using Bt GM crops. A good example of this occurred in China where cotton bollworm (Helicoverpa armigera) became resistant to the pesticides being used in the 1990s and ongoing control came from the use of Bt cotton in the late 1990s (Wu et al. 2008; Lu et al. 2010b). However, Coupe and Capel (2016) indicate that in the USA there has been a substantial reduction in the amount of insecticides used on both maize and cotton since the introduction of GM crops. A study in China has demonstrated that Bt GM cotton was found to be most effective at a low level of pest pressure and that high levels of insecticides have been used despite the cultivation of Bt GM cotton cultivars in response to a decrease in the effectiveness of the technology with an increase in pest pressure (Kuosmanen et al. 2006).

Despite the proposition that the use of glyphosate-tolerant GM crops would reduce (or at least not increase) total herbicide use (Gianessi 2005; Bonny 2008), there is an alternative proposition (Benbrook 2012). It is argued that while in the 5 years after their introduction 1996 lower amounts of herbicides were applied since then, overall herbicide use in herbicide-tolerant GM crops has increased in USA (Schütte et al. 2017). This study quoted Brookes and Barfoot (2015) as evidence of this turnaround. However, Brookes and Barfoot (2015) explain this situation somewhat differently because the non-GM cropping area is so small compared to the GM crop area that the non-GM cropping data set is likely to be biased and unrepresentative of the levels of herbicide actually used. When this bias is addressed by using the average recorded values for herbicide usage on conventional crops for years only when the conventional crop accounted for more than 50% of the total crop and, secondly, in other years (e.g. from 1999 for soybeans, from 2001 for cotton and from 2007 for maize in the US) applying estimates of the likely usage if the whole US crop was no longer using crop biotechnology, based on opinion from extension and industry advisors across USA then cumulatively since 1996, then there have been savings of −3.7% (32.3 million kg) in active ingredient usage and −24.9% for the field Environmental Impact Quotient load. Despite this there are others who believe that herbicide-tolerant GM crops, and in particular soybeans, accumulate herbicides due to overuse of glyphosate (Duke et al. 2003). This has been reviewed by Bøhn and Millstone (2019) who question the amount of glyphosate entering the food chain and their impacts on consumers, which based on other information is mostly used as feed for animals (Lucht 2015; Sieradzki et al. 2021).

Indeed it would seem that average amount of herbicide applied and the associated environmental load, as measured by the Environmental Impact Quotient indicator, have increased on both GM herbicide-tolerant and non-GM crops (Brookes 2014). This has resulted from the increasing incidence of weed species resistance to herbicides and an increased awareness of the consequences of relying on a single or very limited number of herbicides for weed control. Diversifying weed management systems in GM herbicide-tolerant crops has been motivated by the desire to maintain effective weed control with the benefits of reduced tillage. It is still proposed that despite the increase in herbicide use in recent years, GM herbicide-tolerant technology continues to deliver significant economic and environmental gains to US farmers. Others however disagree (Benbrook 2012) or indicate that it depends on the crop. A reduction in annual herbicide application rate has occurred for canola (Gusta et al. 2011) and maize, but has remained unchanged for cotton, and increased for soybean (Coupe and Capel 2016). There is still however the view that GM crops tolerant to herbicides are the main cause for herbicide contaminated water, soils, and wildlife (Cuker et al. 2020), but often with little proof of linked causality.

Secondary pests becoming dominant

The argument is that while GM Bt crops may reduce targeted insect pests, they provide an opportunity for secondary pests to prevail (Zhao et al. 2011). Indeed, it has been shown in USA that Bt GM maize expressing Cry1Ab protein, which controls the European maize borer and maize earworm (Helicoverpa zea), provides a competitive advantage to another pest the western bean cutworm (Striacosta albicosta) particularly when maize earworm numbers and fitness were reduced (Dorhout and Rice 2010). In China, the second country to adopt Bt cotton after the USA, early results showed that the GM cotton had dramatically boosted yield, reduced the impact of cotton bollworms and cut the use of insecticides by as much as 70%, saving money and protecting both people and the environment from the broad-spectrum synthetic toxic chemicals (Pearson 2006; Qiu 2010). However, after 7 years it was observed that populations of other cotton pests, particularly mirids (Apolygus lucorum), had grown significantly in number. This has been confirmed in other studies (Niu et al. 2020) which indicate that feeding on leaves of the GM cotton with either Cry1Ac or CpTI proteins had few toxic effects on mirids. These had previously been controlled by the broad-spectrum insecticides. Future control of these secondary pests would either rely on returning to broad-spectrum insecticides, other forms of biological pest control, or the use of multiple stacked Bt GM cotton cultivars (Wang et al. 2008). Despite this concern a thorough economic analysis using nationally representative data over 15 years has shown that the economic benefit of Bt GM cotton did not diminish but remained stable and continuous in China (Qiao 2015; Qiao and Huang 2020).

Soybean grown in China can be adversely affected by four lepidopteran pests – Spodoptera litura, beet armyworm (Spodoptera exigua), black cutworm (Agrotis ypsilon) and cotton bollworm. Cry1Ac-expressing Bt soybeans may provide good protection against H. armigera but alternative control measures are required to manage the other three pests species (Yu et al. 2013). However, it could be argued that this is not a situation where a secondary pest becomes more dominant but rather indicates the need for a GM soybean with multiple Bt Cry genes that provide a wider pest resistance impact as has occurred in Brazil with multiple stacked Bt genes expressing Cry1A.105, Cry2Ab2, and Cry1Ac (Bacalhau et al. 2020), and in a separate study Cry1Ac and Cry1F (Machado et al. 2020). Experience with Bt cotton has revealed that the emergence of secondary pests requires adjustments to pest management systems to address these ‘new’ pests (Kennedy 2008). However, there is evidence that the adoption of Bt GM cotton did result in pest suppression benefiting non-GM cotton farmers (Wesseler et al. 2011).

Impacts on non-target organisms

GM traits are primarily targeted to control a specific for the pest or pathogen (Rahman et al. 2015), whereas crop protection chemicals may affect beneficial organisms as well as the intended target (Cattaneo et al. 2006). A meta-analysis of 10 publications on Bt GM maize plants expressing resistance to Coleoptera (35%) and Lepidoptera (65%) has shown that GM maize cultivation has no effect on non-target organism in the taxonomic groups Anthocoridae, Aphididae, Araneae, Carabidae, Chrysopidae (adults and larvae), Coccinellidae (adults and larvae), Nabidae, Nitidulidae and Staphylinidae (Pellegrino et al. 2018). However, Braconidae were significantly decreased by 32%, and Cicadellidae were significantly increased. Another meta-analysis of 42 field experiments indicated that non-target invertebrates are generally more abundant in Bt cotton and Bt maize fields than in non-GM fields managed with insecticides (Marvier et al. 2007). A three year field study in Brazil evaluating the effects of Bt GM soybean expressing Cry1Ac and Cry1F proteins showed no effect on the diversity and abundance of non-target arthropods compared with non-GM Bt soybean that was not sprayed with insecticide (Marques et al. 2018). An earlier review also concluded that field studies show that the abundance and activity of parasitoids and predators are similar in Bt and non-Bt crops (Romeis et al. 2006; Chen et al. 2008). In contrast, applications of insecticides on non-GM have usually resulted in negative impacts on biological control organisms. Results comparing single and stacked Bt genes have determined that there is a low probability of harm to non-target organisms so long as the potency of the proteins when combined is not greater than when separate (Raybould et al. 2012). The fate of Bt proteins from transgenic crops has been shown short half-lives through to low level residues lasting for some months (Clark et al. 2005). However, the conclusion was that non-target effects of Bt protein indicate that there is a low level of hazard.

The use of Bt cotton in China has shown that marked increase in abundance of three types of generalist arthropod predators (ladybirds (Coccinellidae), lacewings (Neuroptera), and spiders (Araneae)) and a decreased abundance of aphid (Aphidoidea) pests (Lu et al. 2012). This study also found evidence that the predators might provide additional biocontrol services spilling over from Bt cotton fields onto neighbouring crops (maize, peanut (Arachis hypogaea), and soybean). The use of Bt rice in India has shown no effect on non-target insects (i.e. insects belonging to orders other than diptera and lepidoptera) (Bashir et al. 2004). Similarly, in India the use of Bt cotton has had no impact on cotton aphids (Aphis gossypii) serving as prey or host to several predators and parasitoids and their honeydew is an important energy source for several arthropods (Lawo et al. 2009).

A systematic review of data from studies consisting of 7279 individual invertebrate records from 233 experiments in 120 articles, 75% of which were from peer-reviewed journals published between 1997 and 2020 were used to determine whether growing Bt maize changed the environmental abundance of non-target animals such as arthropods, earthworms and nematodes, especially as compared to growing non-genetically modified maize accompanied by the pesticide necessary to control major pests (Meissle et al. 2022). The review showed that Bt maize had no negative effects on most invertebrate groups including ladybeetles, flower bugs (Insecta), and lacewings. However, populations of Braconidae insects, which are parasitoid wasps that prey on maize borers, were reduced with Bt corn.

An interesting consideration has been the determination of the impact of Bt maize by-products (pollen and decomposing plant litter) on the biology of streams located near to field growing Bt maize (Rosi-Marshall et al. 2007). This study estimated rates of movement of Bt maize by-products into 12 streams and acknowledged that ‘potential impacts of these novel carbon sources could vary depending on the magnitude of the inputs to a given stream’ and that ‘the ultimate impact of these materials on downstream ecosystems is currently unknown and in need of further study’. They did find that 50% of filtering trichopterans collected from the study streams during peak pollen shed had pollen grains in their digestive system, but also that Bt proteins while entering the stream systems ‘did not appear to affect microbial processes associated with litter decomposition’. Yet in laboratory feeding trials they found that ‘the leaf-shredding trichopteran, Lepidostoma liba, had more than 50% lower growth rates when they were fed Bt maize litter compared with non-Bt maize litter (P = 0.008)’, and therefore concluded ‘that pollen adhering to algal biofilms can be consumed by scraping trichopterans and that, at high concentrations, stream dwelling trichopterans can be harmed by the Bt-endotoxin in crop by-products’. One would suggest this was an extrapolation beyond the validity of the data and trial design.

Laboratory studies comparing the effect of two Bt proteins (Cry1Ab and Cry3Bb) expressed in Escherichia coli on the two-spotted ladybird (Adalia bipunctata) (Schmidt et al. 2009). While Cry3Bb showed no consistent impact on larval survival, at 5, 25 and 50 ug/ml of the toxin caused by Cry1Ab there was an increasing mortality up to 35%. However, neither Bt toxin affected larval development time to pupation. Application to a field situation requires these ladybird larvae to be able consume Bt toxins. Exposure of ladybirds to Cry1Ab proteins is unlikely because it is not transported in the phloem of these plants (Raps et al. 2001; Dutton et al. 2002) which is what is accessed by aphids. Therefore, presumably the Cry1Ab toxin maybe contained within aphid feeding on Bt maize, at similar concentrations to those used in the laboratory study. A. bipunctata is a polyphagous ladybird which consumes a range of different aphid species, and it was noted that some aphid species (e.g. the 12-spotted lady beetle – Coleomegilla maculata) can at times consume plant material in the form of pollen, often to overcome food shortages (Lundgren et al. 2005). So the question remains as to whether A. bipunctata can access levels of Bt proteins sufficient high to impact on its mortality, but the authors concluded it cannot be excluded as a possibility. In sampling of beetles in Bt GM maize crops the Cry1Ab protein was identified in two ground beetle species, but the implications of this exposure were unclear (Zwahlen and Andow 2005). Similarly, Bt proteins were found in (non-target) cotton aphids and Propylaea japonica the natural predator of bollworm elder larvae (H. armigera) feeding on Bt GM cotton, but the ecological implications are unclear (Zhang et al. 2004).

GM Maize modified using dehydration-responsive element-binding transcription factors that regulate diverse processes during plant development was grown in a 2-year field study and had negligible influence on arthropod abundance, diversity, and community composition (Yin et al. 2022). Similarly, the effect of herbicide-tolerant GM sugar beet, maize and spring oilseed rape on abundance and diversity of soil-surface-active arthropods showed that direction of the effects was evenly balanced between increases and decreases in counts in the GM herbicide-tolerant crop compared with the conventional treatment (Brookes et al. 2003). For aerial and epigeal arthropods while most were insensitive to the both the herbicide tolerant and the non-GM equivalent the numbers of butterflies in beet and spring oilseed rape and of Heteroptera and bees in beet were smaller under the relevant GM herbicide crop management, whereas the abundance of Collembola was consistently greater in all GM herbicide-tolerant crops (Haughton et al. 2003). They concluded that the results for bees and butterflies might or might not translate into effects on population densities, depending on whether technology adoption leads to forage reductions over large areas. Malone and Burgess (2009) concluded that commercial Bt and herbicide-tolerant GM crops have shown no deleterious effects on pollinators. Other GM crops exhibiting protease inhibitors and lectins may have an effect on bees depending on realistic routes for sufficiently high exposure to occur. Micro-colonies have been used to determine the effect of Bt protein (Cry1Ab), Kunitz soybean trypsin inhibitor (SBTI), or Galanthus nivalis agglutinin (GNA) fed in a sucrose solution to bumble bees (Bombus terrestris) (Babendreier et al. 2008). This showed that while the Cry1Ab did not affect microcolony performance, the consumption of SBTI and especially GNA affected survival of bumble bee workers and drones and caused a significant reduction in the number of offspring.

It appears that most of the concerns about impacts from GM crops on non-target organism, mostly arthropods, are found in laboratory or controlled feeding studies, but rarely in the trial observations. For example, laboratory feeding trials of Bt GM cotton to cotton aphid which is not sensitive to Bt toxin, it was found that trace amounts of Bt toxin was detected in ladybirds preying on Bt-fed aphids (Zhang et al. 2006). The question is – could this effect be extrapolated to a real field situation and if so, would it have any detrimental effect on ladybirds? In one field trial in Australia comparing Bt GM cotton with insecticide sprayed non-GM cotton there were higher numbers of Helicoverpa (as would be expected) and slightly higher numbers of some diptera, damsel bugs (Nabis capsiformis), and jassids (Cicadellidae) in conventional crops (Whitehouse et al. 2005). However, the long term impacts of these differences are unknown.

Using laboratory studies Bt GM crops have been alleged to negatively impact on Monarch butterfly caterpillars (Danaus plexippus) (Losey et al. 1999). However, in reviewing dose–response studies, and exposure characterisation the overall risk posed to the monarch population appears negligible (Sears et al. 2001; Wolt et al. 2003). While exposure of monarch butterfly larvae to Bt pollen is possible (Oberhauser et al. 2001), the significance of that exposure is dependent on several factors involving timing, concentration level of pollen in the air, and rainfall events (Pleasants et al. 2001). It has been concluded that the likelihood of toxicity by Bt maize to monarch butterfly larvae being real and manifested are negligible (Hellmich et al. 2001; Carzoli et al. 2018). However, another study also under laboratory conditions examined the impact of pollen from Bt maize (event Bt-176) on neonate larvae of the Peacock butterfly (Inachis io) which indicated a negative effect on larval weight gain or survival rate (Felke et al. 2010). Larvae feeding on nettle plants (Urtica dioica) with high pollen concentrations from Bt-176 maize (205 and 388 applied pollen cm−2) suffered a significantly higher mortality rate (68 and 85% respectively) compared to larvae feeding on leaves with no pollen (11%) or feeding on leaves with pollen from conventional maize (6–25%). However, at lower Bt maize pollen doses (23–104 applied pollen cm−2), mortality ranged from 11–25% and there were no apparent differences among treatments. The question remains as to whether these laboratory conditions are realistic and would ever occur in a field situation.

Studies on crops with GM traits yet to be commercialised or rarely used have also shown few detrimental effects on non-target organisms. A very recent study (Amin et al. 2022) examining the effect of the human thioredoxin gene under the control of the β-conglycinin promoter with tolerance to the herbicide glyphosate in soybean, showed no negative affect on plant-dwelling non-target insects and arachnids. Another recent study (Oh et al. 2020) also showed no effect on abundance and diversity of plant-dwelling insect/arachnid populations on field grown vitamin A-enhanced GM soybean with tolerance to the herbicide glyphosate. Phytase expressing GM maize had no effect on the relative stability of the arthropod community compared with non-GM maize (Wang and Guan 2020).

Resistant insect populations

Many pests have developed resistance to specific synthetic chemical pesticides (Devine and Furlong 2007; Maino et al. 2018; Hawkins et al. 2019). While there are about 70 types of Cry genes associated with proteins from B. thuringiensis, only a few are used in commercial GM crops (Sanchis 2011; Mehboob-ur-Rahman et al. 2015; Shaheen et al. 2015). It was noted that after the first 8 years of use of Bt crops, despite dire warnings, pest resistance to Bt crops had yet to be documented (Bates et al. 2005). However, field resistance to these proteins produced by Bt Cry genes has since then been observed (Bagla 2010; Huang 2011; Huang et al. 2011; Tabashnik et al. 2013, 2014; Van den Berg et al. 2013; Pandian and Ramesh 2020) and has led to insect resistance management processes (Kebede 2020). A review of 27 data sets involving seven Bt proteins in maize and cotton across six countries found that seven pests had developed severe field-evolved resistance in 2–8 years, eight had statistically significant but less severe resistance, and 12 had no evidence of decreased susceptibility after 2–15 years (Tabashnik and Carriere 2015). It was hypothesised that the increase in pest resistance to Bt since 2005 was associated with the increased planting of Bt crops resulting in increased exposure of pests to Bt proteins and increased monitoring. A global review in 2019 found 19 cases of field-evolved resistance that reduces Bt crop efficacy (Tabashnik and Carriere 2019). Indeed it has been shown that for bollworm that gene frequency for a major gene for Cry2Aa resistance is 1/2584 or 0.039% and for Cry1Acit is 1/2332 or 0.043% (Burd et al. 2003). However, early work demonstrated that stacking Bt Cry expressing genes could provide an option to overcome resistance to a single Cry gene (Zhao et al. 2001).

Evidence of Bt resistance developing in insect pest has led to the use of refuges alongside GM crops (Qaim and De Janvry 2005; Naranjo et al. 2008; Huang et al. 2011; Ives et al. 2011; Van den Berg et al. 2013). The effectiveness of natural refuges depends on the biology of the pest and the distribution and abundance of the host plants in the agricultural system and the quality of the host plants targeted by the pest species (Li et al. 2017). Another viable option is the stacking or pyramiding of Bt genes (Bates et al. 2005; Storer et al. 2012; Van den Berg et al. 2013; Bacalhau et al. 2020). To manage developing resistance to Bt proteins in GM cotton by pink bollworm (Pectinophora gossypiella) two approaches have been successful (Tabashnik and Carriere 2019). In southwestern USA development of resistance was delayed by planting non-Bt cotton refuges from 1996 to 2005, then undertook mass releases of sterile moths, resulting in the eradication of the pest from the region. In China, low levels of pink bollworm resistance to Bt cotton were reversed by planting second-generation hybrid seeds from crosses between Bt and non-Bt cotton. This approach results in a refuge of 25% non-Bt cotton plants randomly interspersed within fields of Bt cotton. In Latin America, only fall armyworm (Spodoptera frugiperda) of the 31 pests targeted and controlled by Bt crops has shown tolerance to certain Bt proteins in growers’ fields (Blanco et al. 2016).

It has been proposed that development of resistance to Bt GM crops can be managed effectively by firstly making sure growers are aware of the potential problem of resistance developing, secondly ensure remedial actions are known (e.g. use of refuges (Huang et al. 2011)) and growers are being compliant (which has been documented as not always being the case (Bourguet et al. 2005; Kruger et al. 2012; Chvatalova 2021)), and thirdly monitor for increases in resistance (Carriere et al. 2020). Seed companies concerned that farmers growing Bt GM maize were not planting sufficient refuge zones led them to introduce a system where a percentage of seed in a bag is non-GM (Yang et al. 2014). This aimed to provide sufficient refuge plants in a maize crop to minimise selective pressure on insect adaptation for resistance.

In a comparison of leaves from insect-resistant single-transgenic cotton (expressing Cry1Ac gene), pyramided-transgenic (expressing Cry1Ac and CPTI genes) and conventional cotton that were consecutively cultivated for three years in the same field showed no difference in leaf area damaged by cotton bollworm collected from the same field (Liu et al. 2020). They hypothesised that this may be due to resistance evolution of the target-insects and decreased Bt content in transgenic cotton leaves.

Increased weediness

GM crop development has been criticised for potentially increased weediness or invasiveness of these plants (Pilson and Prendeville 2004). However, this has not been observed. Early studies investigated the impact of GM crops with herbicide tolerance and pest resistance and concluded that GM arable crops are unlikely to survive for long outside cultivation (Crawley et al. 2001). The comparison of herbicide–tolerant beet, maize and spring oilseed rape on 12 weed species compared with non-GM equivalent crops showed for many weed species in beet and spring oilseed rape that both weed density in the crop and seed densities were lower in the seedbank after herbicide-tolerant GM cropping, but in maize these weed populations may increase (Heard et al. 2003a, 2003b).

Impacts on biodiversity

It has been stated that ‘favouring biodiversity does not exclude any future biotechnological contributions, but favouring biotechnology threatens future biodiversity resources’ (Jacobsen et al. 2013). They proposed that research should be focused on areas of plant science, e.g. nutrition, policy research, governance, and solutions close to local market conditions if the goal is to provide sufficient food for the world’s growing population in a sustainable way. Much of the concern about using GM crops was associated with the domination of multinational companies and placing farmers into a position where they are reliant on the agro-industry, that GM crop development is just as slow as cultivar development using conventional breeding techniques, that extensive systems are more sustainable than intensive systems of farming, and the need for regulatory hurdles to be negotiated before GM crops can be released. It is always possible to find examples where technologies used in agricultural production have a possible negative or less than desirable effect. However, that does not automatically render the technology redundant but rather indicates improvements in use are required through better management.

Deforestation of the Amazon, and hence loss of biodiversity, began prior to the use of GM soybeans but has increased with the expansion of GM soybean cropping in Brazil (Altieri and Pengue 2006; Elferink et al. 2007), although deforestation due to pasture expansion is also a significant factor (de Waroux et al. 2019). However, while Brazil is a major producer of GM soybean and most of the increase in production there has been driven by increased demand for soybean use for animal feed, biofuels and vegetable oil (Ritchie and Roser 2021) it has been concluded that the dominant driver of deforestation in the Brazilian Amazon was the expansion of pasture for beef production (63% of total loss area) and then small-scale forest clearing (12%) (Barona et al. 2010; Tyukavina et al. 2017).

Biodiversity has been promoted as a necessary requirement for beneficial ecological outcomes (Tilman et al. 2014; Schütte et al. 2017). There is no doubt that preservation of biodiversity in natural ecosystems is of paramount importance, however, there is still debate on whether biodiversity is beneficial or even necessary in productive agricultural ecosystems. Modern agricultural practices using synthetic chemical pesticides have been alleged in reducing biodiversity (Devine and Furlong 2007). It could be considered that biodiversity in productive pastoral ecosystems might be beneficial, but it has been shown that the addition of forbs (i.e. herbaceous flowering plants that are not grasses) that contain plant-specialised metabolites to a grass–white clover mixture did not, under the conditions of the experiment reported, provide any additional benefits (Loza et al. 2021). The ecological production function which quantifies the relationship between ecological inputs and particular ecosystem outputs has been modelled using data from the longest-running biodiversity experiment in the world, at Cedar Creek Ecosystem Science Reserve, in USA (Binder et al. 2018). They found that even a risk-neutral, profit-maximizing landowner would favour a highly, but not maximally, diverse mix of 11 species. However, a negative relationship was observed between productivity and diversity in the annual grasslands of California and the old fields of New York, while there was no relationship between productivity and diversity was found in the Serengeti of Kenya (McNaughton 1993). Yet it has been observed that the effect of perturbation on production was maximised in simple systems and minimised in the most diverse systems (Tilman and Downing 1994).

Modelling has been used to understand the impact of GM herbicide-tolerant crop technologies on seed eating birds and not surprisingly concluded that when weed populations are reduced to low levels or practically eradicated, consequent effects on the local use of fields by birds might be severe, because such reductions represent a major loss of food resources (Watkinson et al. 2000). According to Dewar et al. (2005) the ban on use of GM crops in the UK was based on results from 60 to 70 field trials of herbicide-tolerant brassicas and maize which showed in 3 of the 4 crop comparisons a reduction in weed diversity using GM crops compared with conventional non-GM crop management. So for being too efficient in weed management the GM crops were banned, resulting in the withdrawal of biotech investment in R&D from Europe. However, this could equally apply to weed control measures used for non-GM crops.

An interesting analysis of the EU Farm to Fork (F2F) strategy has indicated a decline in EU agricultural production in quantitative terms. This has been justified by policy makers on the basis that the additional net benefits outweigh the losses in consumer surplus (Wesseler 2022). However, the unintended consequence of a decline in EU agricultural production and a ban on all soybean imports (Henning and Witzke 2021), most of which are GM soybean (Shahbandeh 2022), is that land use change outside of the EU may increase, driving deforestation and reducing biodiversity. Indeed it has been estimated that without plant breeding in the EU driving up gross domestic product in the last 20 years, an extra 8.3 million hectares of rainforest and savannahs in Brazil or 11.8 million hectares of natural habitats in Indonesia would have been lost (Noleppa and Cartsburg 2021).

Compared with conventional pest control measures, which can rely on broad-spectrum insecticides the majority of studies involving GM crops have not found any profoundly negative effects on arthropod natural enemies (O'Callaghan et al. 2005). Devine and Furlong (2007) surmised that the negative effects that have been reported due to the use of GM crops have generally been subtle and rather difficult to predict. As an example, the mortality of green lacewing larvae (Chrysoperla carnea) increased when fed on moth larvae (Spodoptera littoralis) reared on Bt maize but not when fed on spider mites (Tetranychidae) reared on Bt maize (Hilbeck et al. 1998; Dutton et al. 2002). However, in the field environment lacewing larvae did not chose moth larvae as a first choice food (Meier and Hilbeck 2001), so this concern may not in actuality be real. The point here is that while often (contrived) laboratory experiments raise concerns the more thorough long-term field trials are rarely carried out to assess real effects of exposure to a hazard or fitness over several generations (Gray 2004). In a comprehensive review it was concluded that the predator green lacewing was not affected by Bt GM maize or by the produced Cry1Ab protein (Romeis et al. 2004, 2014). Similarly, tests with diets containing Cry1Ab, Cry1Ac, Cry1Ah, Cry1Ca, Cry1F, Cry2Aa, Cry2Ab, and Vip3Aa showed no effects on another predatory green lacewing species (Chrysopa pallens) larvae (Ali et al. 2018). However, these debates on potentially unintended consequences towards invertebrates is one that is contentious (Hilbeck et al. 2012).

A study to determine the diversity and abundance of predatory arthropods (insect and spiders) in Bt and non-Bt cotton fields in Telangana state, India found 9 families of predatory insects and 2 families of spiders on Bt cotton and 11 families of predatory insects and 3 families of spiders on non-Bt cotton (Mallesh and Sravanthy 2021). There were 14 species of predatory arthropods on Bt cotton and 17 species on non-Bt cotton, 13 of which were common between the two groups. They concluded that Bt cotton may affect predatory arthropods indirectly. An ecological study in China spanning 2 decades and across 36 sites clearly showed that with the commercial uptake of Bt cotton in 1997 there was a significant increase in predatory arthropods including ladybirds, spiders, and lacewings (Lu et al. 2012).

The use of GM crops per se has been argued as a threat to biodiversity within crop species (Gepts and Papa 2003). However, several studies have concluded that the use of GM crops has not significantly affected levels of genetic diversity within crop species (Bowman et al. 2003; Sneller 2003; Ammann 2005).

Transgene escape into wild or non-GM populations

Gene flow from GM crops into wild or weedy relatives has been viewed a potential unintended consequence (van de Wiel et al. 2003; Poppy 2004; Lu and Snow 2005; Wilkinson and Ford 2007; Lu 2008; Ryffel 2014) resulting in their increased fitness through improved resistance to insects, diseases, herbicides, or harsh growing conditions (Snow 2002), or genetic erosion of commonly owned landraces (Garcia and Altieri 2005). It has been noted that many crop plants have wild relatives with which they can hybridise (Ellstrand et al. 1999; Cruz-Reyes et al. 2015). However, one viewpoint is that the likelihood of vertical gene transfer from GM crops to different species is not different from that for other plant DNA (De Santis et al. 2018; Niraula and Fondong 2021).

One of the first commercialised GM crops was CDC Triffid in 1998, a herbicide-tolerant linseed/flax (Linum usitatissimum) cultivar developed in Canada (Ludvikova and Griga 2015). But it was deregistered in 2001 and removed from the market at the request of the Flax Council of Canada and the Saskatchewan Flax Development Commission, as a reaction to the EU’s concern with importing linseed contaminated with GM seed, through cross pollination. However, later that decade Triffid flax was unexpectedly detected in EU food products and in subsequent flax imports from Canada to Europe. This has resulted in all exports of flaxseed from Canada now having to be screened for nil GM contamination.

For out-crossing species, such as brassica (Warwick et al. 2003; Ford et al. 2006) it is difficult, if not impossible, to prevent gene flow from GM cultivars to wild/weedy and non-GM plants of the same or related species. The development of herbicide resistant creeping bentgrass (Agrostis stolonifera) demonstrated the unintended ecological consequence of transgene introgression into wild populations (Watrud et al. 2004; Reichman et al. 2006) and led to it being withdrawn from the market. Pastoral agriculture in New Zealand is dominated by wind pollinated grasses for which gene transfer has been shown to occur to a distance of 150 m for tall fescue (Festuca arundinacea) (Wang et al. 2004), 80 m for perennial ryegrass (Lolium perenne) (Giddings et al. 1997), and 250 m for meadow fescue (Festuca pratensis) (Rognli et al. 2000). A leptokurtic distribution for gene flow where differences in the rate of decline over distance depends on wind direction (Cunliffe et al. 2004). The same would also apply to outcrossing pastoral legumes (Woodfield et al. 1995; De Lucas et al. 2012). However, the commercialisation of Roundup Ready lucerne demonstrates that regulatory approval can be obtained for a perennial outcrossing forage legume (Wang and Brummer 2012).

The question then to consider is whether the transfer of transgene would be detrimental and create an unmanageable risk? It has been concluded that while transgenes that confer resistance to pests and environmental stress and/or lead to greater seed production have the greatest likelihood of aiding weeds or harming non-target species (Légère 2005; James 2006) this is unlikely for most currently grown transgenic crops (Prakash et al. 2011). Hybridisation between male sterile GM canola (Brassica napus) and B. rapa in the UK appears inevitable (Wilkinson et al. 2003; Légère 2005). However, it was concluded that ‘the presence of hybrids is not a hazard in itself and does not imply inevitable ecological change’. Despite that the glyphosate-tolerant trait persisted in wild populations of brassica over a 6-year period even in the absence of herbicide selection pressure and in spite of the fitness cost associated with hybridisation (Warwick et al. 2008).

Pollen drift may be prevented by carefully maintaining crop-specific isolation distances and using border strips around fields to trap pollen or by some molecular methods (Kim et al. 2020; Prabha et al. 2020). Selective advantage and fitness of a trait associated with a transgene in a GM crop is an important consideration such that traits related to the success of gene flow, resistance to biotic or abiotic stress may result in selective advantages or serious fitness improvements (Ammann et al. 1994).

Impact on rhizosphere microorganisms

A review on direct, indirect, and pleiotropic effects of GM plants on soil microbiota showed impacts that depend on transformation events, experimental conditions and taxa analysed (Turrini et al. 2015) (refer to Table 7 for a summary). This review showed that 32% of comparisons indicated a consistent effect of GM technologies on soil microorganisms. They were however unable to determine whether the effects of a specific transgenic crop are clearly beyond the differences that would be found between a range of non-GM cultivars. They concluded that adequately designed and standardised tests are required to assess the impact of transgenic crops on soil microorganisms, particularly those important soil ecological functions, e.g. nitrification, nitrogen fixation, phosphate mobilisation, organic carbon cycling and sink. Studies reviewing the impact of synthetic pesticides, fungicides, and herbicides on soil microorganisms have shown potential disruption of N cycling, which can be soil type and dose dependent (Che et al. 2022; Sim et al. 2022) and C source (Streletskii et al. 2022). The impact of Bt11 and Bt176 GM maize plants and their residues on bulk soil and rhizospheric eubacterial communities on the arbuscular mycorrhizal fungus Glomus mosseae and on soil respiration was tested using microcosms (Castaldini et al. 2005). Differences in rhizosphere eubacterial communities were apparent, and there was a significantly lower level of mycorrhizal colonisation on in Bt176 maize roots. For Roundup Ready soybean (event MON 89799) there were no significant differences compared with the control non-GM cultivar (A3244) for nodule number, shoot total nitrogen, and dry weight of nodules (Horak et al. 2015). The question of importance here is whether the effect of GM crops on soil microorganisms and ecosystems is more or less than that caused using synthetic chemistry?

Table 7. Number of references indicating an impact or not of transgenic events on soil microorganisms due to pleiotropic and undetermined (pleiotropic or direct) effects. Taken from review by Turrini et al. (2015).

Transgenic trait or gene	Effects of transgenes expressed in different plant species on soil microorganisms
	Consistent impact	Transient impact	No effect
Bt genes for insect resistance	19	13	35
Resistance to pathogens	6	7	8
Tolerance to herbicides	6	5	8
Food nutritional benefits	9	3	6
Total	40	28	57

Interestingly, the insertion of the Cry1Ab gene rice had no significant effect on the residual decay and decomposition-associated microbial community compositions in a canola-rice cropping system (Lu et al. 2010a) and paddy rice system (Wu et al. 2009). This may have been due to the rice expressing the Cry1Ab gene having no measurable adverse effect on the key microbial processes or microbial community composition in rhizosphere soil over 2 years of rice cropping (Liu et al. 2008; Wei et al. 2012). In pot trials Bt11 GM maize had similar G. mosseae colonisation at high fertiliser rates with the non-GM control, but lower colonisation at low fertiliser rates (Cheeke et al. 2011). In an extensive review Mandal et al. (2020) concluded that while some studies showed that GM crops caused considerable changes in the structure and functions of indigenous soil microbial community, interpreting the real impact of GM crops on soil microorganisms was often confounded by the soil heterogeneity, varying nutritional requirements crops and the lack of suitable controls. Across 25 studies in that review only four indicated some impact on soil microbial communities, and of those only two suggested it was negative impact. The long-term effects, if any, of these impacts on soil microorganisms is yet to be understood. In field trials of GM potatoes able to express lectins to control invertebrate pests any rhizosphere effect did not persist and had no effect on the subsequent barley (Hordeum vulgare) crops (Griffiths et al. 2000). Likewise actively growing Bt GM maize and soil incorporated plant straw had no constant apparent effect on the soil bacteria and fungi community structure (Tan et al. 2010). In India, the use of Bt GM cotton again had not adversely effects on the diversity of the soil microbial communities (Kapur et al. 2010). In Argentina, a GM drought tolerant maize cultivar expressing the Hahb-4 gene did not have any undesirable impacts on rhizosphere bacterial communities (Ibarra et al. 2020). Despite this overwhelming evidence there are still calls for laboratory testing to determine the influence of GM crops on plant-associated, soil and environmental-associated microbiota (Pepoyan and Chikindas 2020).

The effect of Bt proteins on soil invertebrates has indicted few significant impacts (some of which have been positive) except on soil nematodes (Icoz and Stotzky 2008) (Table 8). They established that Cry proteins released in root exudates and from plant residues of Bt GM crops appear to have no consistent, significant, and long-term effects on the microbiota and their activities in soil. This led to the conclusion that with few or no significant detrimental effects of Cry proteins on microbes and other organisms in below-ground soil ecosystems that the low level of risk associated with Bt GM crops ongoing risk analysis studies are probably not warranted.

Table 8. The impacts of Bt proteins administered in the laboratory or measured in the field on soil invertebrates. Summarised from Icoz and Stotzky (2008).

Organism group	Number of studies	Study location*	Effect of cry proteins on organism
			Consistent impact	No effect
Earthworms	6	0.17	0	6
Woodlouse	4	0	1 (fed less)	3
Collembola	9	0.33	1 (changed feeding preference)	8
Nematodes	12	0.25	7	5

*Proportion of total studies undertaken that were in the field.

Horizontal gene flow

Horizontal or lateral gene transfer is the stable transfer of genetic material from one organism to another with reproduction or human intervention (Keeling and Palmer 2008; Keese 2008). Indeed gene transfer processes between bacteria in the phytosphere may be part of their evolutionary development and adaptation to plant rhizospheres (van Elsas et al. 2003). However, it has been concluded that the frequency of horizontal gene transfer from plants to other eukaryotes or prokaryotes is extremely low, but to viruses it is potentially greater but the impact is restricted by selection pressures. Keese (2008) in a thorough review concluded that horizontal gene transfer from GM plants poses negligible risks to human health or the environment.

Using the example of the development of penicillin-resistant Streptococcus pneumoniae as an example of how gene transfer can occur it has been argued that when genes evolve by transfer rather than through organismal reproduction, neither the generation time nor the geographical range of the organism necessarily limits the lag time (Heinemann and Traavik 2004). They argue that horizontal gene transfer can occur from crops to soil microbes, and this will have ‘an environmental impact at a frequency approximately a trillion times lower than the current risk assessment literature estimates the frequency to be’. They believe that ‘ current methods of environmental sampling to capture genes or traits in a recombinant are too insensitive for monitoring evolution by horizontal gene transfer’ and propose ‘a model involving iterative short-patch events explains how horizontal gene transfer can occur at high frequencies but be detected at extremely low frequencies’. More recent theory suggests that here is little evidence of horizontal gene transfer being a major process in determining resistance frequencies in S. pneumoniae but are best explained as the outcome of selection acting on a pool of variants, irrespective of the rate at which resistance determinants move between genetic lineages (Lehtinen et al. 2020).

A review of possible ecological effects of GM plants on microbes in the field, both at the plant level (the phytosphere) and in the soil matrix concluded that must have occurred in the past due to the similarity between genes from phylogenetically remote organisms that could only be the result of horizontal gene transfer (Pontiroli et al. 2007). DNA can persist in soil under certain circumstances and under optimised conditions in vitro recombination between eukaryotic DNA and prokaryotic DNA has been observed (Gebhard and Smalla 1998) but has not been documented in fields planted with GM crops (Gebhard and Smalla 1999). While still a potential biosafety issue the chances of increasing the fitness of any bacteria acquiring the genes from a GM plant is remote (Thomson 2001) and it been argued that the focus should be on the real causes of bacterial resistance such as the continued overuse of antibiotics by physicians and veterinarians (Salyers 1996; Smalla et al. 2000).

The argument that CaMV 35S is only plant specific and not active in other organisms such as bacteria and fungi has been challenged with some cause (Steinbrecher 2002). There is no documented evidence of the transfer of CaMV 35s into microbes from GM crops in the field (Bak and Emerson 2020), although it has been demonstrated as possible, but only under very specific conditions, such as strong selection pressure in the laboratory (Wintermantel and Schoelz 1996).

Using expressed sequence tag analysis one gene designated ShContig9483, which shows high similarity to genes in sorghum (Sorghum bicolor) and rice suggested horizontal gene transfer must have occurred between the parasitic plant Striga hermonthica and the infected monocot host plant (Yoshida et al. 2010). This indicates that while potentially possible the chance of transgenes from rice moving to the parasitic plant S. hermonthica is very remote.

Increased antibiotic resistance

If transgenes can move through horizontal gene transfer, then one concern would be the rise of antibiotic resistance. The production of GM plants most often uses a genetic construct which includes not only the gene of interest and the relevant promoter, but also an antibiotic-resistant gene, used as a selectable marker. This has raised concerns of these antibiotic-resistant genes either directly affecting native soil bacteria or even creating antibiotic-resistant microorganisms through horizontal gene transfer (Turrini et al. 2015). Transfer between microorganisms of plasmids with an antibiotic resistance gene has been reported in laboratory soil chambers (van Elsas et al. 1988). The horizontal transfer of kanamycin resistant genes used in GM plants to the rhizosphere bacterium Acinetobacter sp. has been demonstrated in sterile soil microcosms (Neilsen et al. 2000). However, Neilsen et al. (2000) concluded that while this may take place in soil, the environmental significance of such rare events depends upon selection for the acquired character. Likewise, Bennett et al. (2004b) conclude that the risk of transfer of antibiotic resistance genes from GM plants to bacteria is remote (Bennett et al. 2004b), and that the hazard arising from any such gene transfer, incorporation, and transmission to humans is extremely remote (Gay and Gillespie 2005). After over 25 years of commercial use of GM crops there is no documented evidence of this occurring. Additionally, modern techniques for genetic modification include the Cre recombinase gene excision system, where the antibiotic resistance selectable marker can be excised after regeneration so that the final GM plant lacks the antibiotic resistance gene (Thomson et al. 2009; Tuteja et al. 2012). This can be routinely used in a range of crops (Zhang et al. 2022).

Gene transfer to consumers from GM food and feed

The transfer of transgenic DNA from feed to tissues of piglets was considered as a natural process and that the risk of gene transfer from GM food is no different from gene transfer in non-GM feeds (Mazza et al. 2005). The movement of the 5-enolpyruvylshikimate-3-phosphate synthase gene found in GM soybean to humans found that while it might survive through into the small intestine it was completely degraded in the large intestine and did not alter gastrointestinal function nor pose a risk to human health (Netherwood et al. 2004). Polymerase chain reaction analyses of milk samples collected from dairy cows fed Bt GM maize silage showed neither tDNA (event T25) nor the single-copy endogenous maize gene, alcohol dehydrogenase (Phipps et al. 2005). Additionally, ELISA assays did not detect the protein expressed by the T25 gene in milk. In another study, no milk samples from cows fed GM maize or soybean were found to contain recombinant DNA within the limit of detection (De Giacomo et al. 2016). This means that neither transfer of genetic material nor aerosol contamination from feed to milk was shown. Pigs fed GM soybean meal (Roundup Ready MON 40-3-2) and Bt maize (MON 810) had detectable amounts of transgenic DNA in the stomach and duodenum but not in the digesta of further intestine parts, faeces, blood, and evaluated organs and muscles (Świątkiewicz et al. 2013).

Unintended compounds produced

The concern of either new or known toxins being produced by GM crops has been proposed (Kessler et al. 1992), but with very little evidence (Ladics et al. 2015). Conversely, it has been shown that Bt GM maize not only does not produce unexpected toxins but may reduce fumonisins, deoxynivalenol and zearalenone contamination (mycotoxins) as a health benefit (Ostry et al. 2010). Reduced fungal infection and hence potential production of mycotoxins, has been demonstrated due to reduced damage from European maize borers (Munkvold et al. 1997, 1999). The use of Cauliflower mosaic virus 35S (CaMV 35S) promoter has been examined for possible unintended consequences. No relevant similarity was identified between the putative peptides and known allergens and toxins due to the use of CaMV 35S promoters, although long variants may contain an open reading frame, which when expressed might result in unintended phenotypic changes (Podevin and du Jardin 2012). Instability of transgenes leading to antinutritional unintended consequences has been hypothesised despite acknowledging that consumption of GM plant-based food is as risky as (or alternatively as safe as) consuming conventional plant-based food (Ogwu 2021).

Summary – balancing the risk and benefits

Despite all the warnings and fearmongering about the perils of using GM crops it has been concluded by many after 20 years of their use that most of the risks associated with the use of GM crops have proven to be low to non-existent (Carzoli et al. 2018; Vega Rodríguez et al. 2022). Perhaps the most concerning is the possible development of GM-induced insect resistance or plant herbicide resistance, but such resistance is not confined to GM crops as resistance in target insects and weeds is also evident in non-GM management systems (Mannion and Morse 2013). Even though GM crops have been used for either animal feed or human food for over 25 years in USA there has been no legitimately recorded cases of health-related issues. All concerns linking health-related issues to GM feed fed to animals and possibly extrapolated similar effects to humans have come from laboratory trials. Despite that there is still the view that just because it has not been observed that does not mean to say it has not happened or will not happen (Carman 2004). This is an untestable position being used to discredit GM crops and create fearmongering. However, it is still of paramount importance to ensure that new GM crops are tested for their safety as they enter the food chain.

Prevailing negative commentary on GM crops must be countered by the provision of reliable, fact/data based and peer reviewed information that spells out the benefits and how risks, if any are managed, for better outcomes for the environment, economy, and society (Reiss 2001; Arntzen et al. 2003; Burke 2004; Cook et al. 2004; Gaskell et al. 2004; Nerlich et al. 2004; Ghasemi et al. 2013; Huesing et al. 2016; Kotey et al. 2016; Tallapragada et al. 2020; Kubisz et al. 2021). GM crops with a unique or novel trait are each a discrete technology and need to be discussed on its merits rather than simply as a crop developed through genetic modification (Arntzen et al. 2003). Rational public debate and dialogue on the benefits and risks of GM crops is needed (Borch and Rasmussen 2002, 2005; de Bakker et al. 2016). These need to be open and fact based.

Coexistence – likely, possible or creates more issues?

Co-existence of GM crops with non-GM crops is an area of contention and debate (Belcher et al. 2005; Messean et al. 2006; Levidow and Boschert 2008; Devos et al. 2009; Kalaitzandonakes et al. 2016). In a survey of UK farmers on their attitude towards using GM crops (namely maize, canola and sugar beet) if they were permitted the question of how to manage co-existence with non-GM crops was considered (Jones and Tranter 2014). Of potential actions that might be required to manage co-existence the least burdensome was keeping records of seed purchases and product sales for five years. The most burdensome measure was seen as planning crop sowing in such a way that would not coincide with a neighbour’s planting which would result in additional cost due to sowing a 12-row buffer zone around the GM crop. Drivers of GM technology uptake include improved net income and flexibility in crop management (Marra and Piggott 2006), both of which may be restrained by methods needed to allow for acceptable co-existence. Indeed, the implementation of a coexistence policy is considered to have negative impact on farmers’ attitudes towards adoption of GM crops (Areal et al. 2011). If the cost of achieving coexistence between GM and non-GM crops measures falls exclusively on GM producers that is likely to lead to a competitive equilibrium that is biased against GM products (relative to the welfare maximising allocation (Moschini 2015)).

Managing gene flow from an outcrossing species is inevitably challenging if not impossible. One view is that escape of transgenic pollen is inevitable, and therefore the focus should be more on risk analysis about the likely ‘invasiveness’ of the transgenic plants and any ‘mitigation’ options available to reduce their impact on natural, as well as agricultural systems (Kareiva et al. 1994). Estimates of likely cross-fertilisation of major GM crops at different isolation distances have been summarised by Giraldo et al. (2019). To achieve cross-fertilisation levels below 1% an isolation distance of 20 m is required for maize, 9 m or more for cotton, and 5 m for soybean. Co-existence strategies for outcrossing forages have been reviewed (Baltazar et al. 2015; Smith and Spangenberg 2016). Other useful values for pollen detection at distance for a range of commercialised crops is provided by Christey and Woodfield (2001).

The GM tolerance threshold of seed imported into New Zealand is 0% (MPI 2019), not 1% as quoted by Giraldo et al. (2019). The 1% refers to the permitted level of GM ingredient in unlabelled food so long as it was unintentional (MPI 2021). In Australia, South Africa, Brazil, and China 1% adventitious GM seed is permitted, while it is 5% in US, Canada, and Japan, and 0.9% in European Union (Schenkelaars and Wesseler 2016; Giraldo et al. 2019). In North America, emphasis is on the balance between GM and non-GM crops, the isolation distances are based on scientific principles and both GM and non-GM farmers have a stake in preventing adventitious presence (Ramessar et al. 2010). The regulated 0% tolerance is undoubtedly a barrier to adoption of GM technologies in New Zealand.

Farm-level coexistence policies vary markedly between countries, even within the EU where only one country, Spain has permitted cultivation of a GM crop, maize (Beckmann et al. 2006; Schenkelaars and Wesseler 2016). Interestingly, Spain requires the lowest isolation distance for maize in the EU – and it seems to work effectively. In Europe, co-existence is regulated through ex-ante (‘before the event’) regulation and ex-post liability rules (Beckmann et al. 2010). This can create additional costs to the adopters, requirements for a minimum distance between GM and non-GM crops, with the latter impacting on farm size where GM crops might be grown (Beckmann et al. 2010). Further analysis has shown that strict liability will deter the cultivation of Bt GM maize in Germany unless these issues can be addressed through other means, for example, through workable agreements with neighbours (Venus et al. 2017). Informing neighbours and keeping within minimum distances can deliver higher gross margins from growing GM maize rather than non-GM maize (Skevas et al. 2010).

Future opportunities

While initial developments using GM technology focus on input traits future opportunists are more likely to include output traits (also called second generation GM crops) that will benefit consumers (Napier et al. 2019a; Sheoran et al. 2022) which may result in a broad acceptance of GM crops (Giannakas and Yiannaka 2008).

Rice is an important food crop, second only to maize, with a global harvest area of 158 million ha and annual production of 700 million tons (Shaheen et al. 2022). However, GM rice has been slow to be commercialised (Brookes and Barfoot 2003; Jin et al. 2019), despite demonstrated benefits that improve pest resistance, nutritive value, and drought tolerance (Table 9). Despite the continued overuse of pesticides on rice (Damalas and Eleftherohorinos 2011; Zhang et al. 2015; Huang et al. 2022), because pests of rice cause significant losses (Lu et al. 2018), Bt rice is not being grown in any country (Jin et al. 2019), although it has been approved for imports and consumption by the US Food and Drug Administration (FDA) (2018) (Jin and Drabik 2022).

Table 9. Representative examples of potential benefits of genetic modification to crop plants – unexploited opportunities and future options – (also refer to Newell-McGloughlin 2008, McGloughlin 2010; Nalluri and Karri 2020 for further detail).

Trait	Crop	Intended consequence; efficacy and benefits	Reference
Herbicide tolerance	Rice	Engineering herbicide-resistant rice plants through CRISPR/Cas9-mediated homologous recombination of acetolactate synthase	Sun et al. (2016)
Insect pest resistance due to Bt Cry proteins	Broccoli (Brassica oleracea var. italica)	Pyramided Cry1Ac and Cry1C Bt genes control diamondback moth (Plutella xylostella) even those resistant to Cry1A or Cry1C toxins	Cao et al. (2002)
	Eggplant, Brinjal	Eggplant engineered to produce the Cry1Ac protein provided control of eggplant fruit and shoot borer (Leucinodes orbonalis) reducing need for pesticides	Hautea et al. (2016)
	Pigeon pea (Cajanus cajan)	Marker free transgenic events with Cry1Ac protein were found to confer resistant activity against pod borer Helicoverpa armigera larvae	Ganguly et al. (2022)
	Rice	Confers resistance to stem borers (Scirpophaga incertulas and Chilo suppressalis) and leaf folders; no direct negative effects on predators, parasitoids, non-lepidopterous herbivores, or soil invertebrates and soil health. But not commercialised by 2016.	Cohen et al. (2008); Fraiture et al. (2016)
	Table potatoes	Bt table potatoes were produced from 1996 to 2001 for control of Colorado potato beetle (Leptinotarsa decemlineata), due to complications connected with planting GM potatoes, new insecticide compounds, and rejection of the public	Grafius and Douches (2008); Kadoić Balaško et al. (2020)
Insect pest resistance	Cocoa (Theobroma cacao)	CRISPR/Cas9 used to insert the Non-Expressor of Pathogenesis-Related 3 (TcNPR3) gene, a suppressor of the defence response; resulted in increased resistance to infection with the cacao pathogen Phytophthora tropicalis	Fister et al. (2018)
	Rice	Plants expressing the cecropin A gene from the giant silk moth Hyalophora cecropia show enhanced resistance to Magnaporthe grisea, the causal agent of the rice blast disease	Coca et al. (2006)
	Various crops	Alternative GM technologies using lectins, protease inhibitors; multiple resistance genes; tissue specific expression	Christou et al. (2006)
Drought tolerance	Maize	ARGOS8 gene (a negative regulator of ethylene responses) variants of maize increased grain yield by five bushels per acre under flowering stress conditions and had no yield loss under well-watered conditions.	Shi et al. (2017)
	Multiple species	Expression of related cold shock proteins (CSPs) from bacteria, CspA from Escherichia coli and CspB from Bacillus subtilis, promotes stress adaptation	Castiglioni et al. (2008)
		Pathways and genes that can be manipulated through biotechnology to improve drought tolerance.	Deikman et al. (2012)
	Rice	GM rice containing the CaMsrB2 gene are superior in performance to non-GM rice in drought stress conditions	Racheal et al. (2020)
		Overexpression of stress responsive gene SNAC1 (STRESS-RESPONSIVE NAC 1) significantly enhances drought tolerance in transgenic rice	Hu et al. (2006)
		DREB 2A gene isolated from drought tolerant rice variety	Raj et al. (2015)
		Introduction of ZAT12 gene improved drought tolerance in rice	Devi et al. (2016)
	Wheat	NAC gene, TaNAC5D-2 a positive regulator of drought tolerance when silence in wheat increased water loss and decreased growth compared to control and negative control plants under drought stress	Ma et al. (2022)
Salinity tolerance	Chickpea (Cicer arietinum)	OsRuvB gene used to improve salinity tolerance	Preeti (2022)
	Goundnut (Arachis hypogaea)	Overexpression of AtNHX1 gene not only improved salt tolerance but also drought tolerance in transgenic groundnut	Asif et al. (2011)
	Mulberry (Morus indica)	Improved salinity tolerance due to the transgene barley hva1 gene which encodes a group 3 LEA protein and is induced by ABA and water deficit conditions	Lal et al. (2008)
	Rice	Overexpression of stress responsive gene SNAC1 (STRESS-RESPONSIVE NAC 1) significantly enhances salinity tolerance in transgenic rice	Hu et al. (2006)
		Over expression of glyoxalase II that increased further upon salt stress, reflecting the upregulation of endogenous glyoxalase II.	Singla-Pareek et al. (2008)
		At the seedling stage, the salinity tolerance of T2 homozygous mutant lines with the OsRR22 gene was significantly enhanced compared to wild-type plants	Zhang et al. (2019)
	Tobacco (Nicotiana tabacum)	Three salt-stress induced hemp NAC genes, especially CsNAC1 and CsNAC3, played significant roles in the improvement of salt tolerance in transgenic tobacco	Liu et al. (2022)
	Tomato	Improved tolerance to salinity	Zhang and Blumwald (2011)
Nutritional and health benefits (also refer Garg et al. 2018)	Banana (Musa spp.)	6-fold increase in iron accumulation and a 4.5-fold increase in the zinc levels were showed in transgenic banana expressing ferritin constitutively	Chen et al. (2014)
		The fifth exon of the lycopene epsilon-cyclase (LCYϵ) gene was targeted to develop the β-carotene-enriched Cavendish banana cultivar	Kaur et al. (2020)
		Increased level of provitamin A	Namanya et al. (2020)
	Barley	Cisgenic lines with a single extra copy of the HvPAPhy resulted in 2.6- and 2.8-fold increases in phytase activities, that should result in increasing bioavailable phosphate from the stored phytic acid	Holme et al. (2012)
	Brassica	Modification of a fatty acid desaturase 2 gene (FAD2) leading to an increase in the content of oleic acid compared with that in wild-type seeds.	Okuzaki et al. (2018)
		Expression of omega-3 oils – eicosapentaenoic acid (EPA) and docosahexanoic acid (DHA) – in canola	Napier et al. (2019b)
	Camelina sativa	Expression of omega-3 oils – eicosapentaenoic acid (EPA) and docosahexanoic acid (DHA)	Usher et al. (2015)
	Cassava	Modification of starch quality in storage roots	Bull et al. (2018)
	Cereals	Express additional proteins which are rich in lysine and/or methionine	Shewry (2007); Galili and Amir (2013)
	Maize	Use of zinc-finger to alter the inositol phosphate profile in developing seeds	Shukla et al. (2009)
	Peanut	Positive AhFAD2 mutant lines displayed a 0.5–2 fold increase in the oleic acid content compared with non-transgenic controls	Wen et al. (2018)
	Potato	Silencing two asparagine synthetase genes through ‘all-native DNA’ Transformation of potato. French fries and potato chips accumulated as little as 5% of the acrylamide (toxic and carcinogenic – Perera et al. 2021) present in wild-type controls.	Rommens et al. (2008)
	Rice	High expression of amylose (resistant starch) rice through CRISPR/Cas9-mediated editing of SBEIIb, to reduce risk of non-infectious diseases	Sun et al. (2017)
		Knock out of Inositol 1,3,4-trisphosphate 5/6-kinase6 (IPTK6) gene results in low phytic acid expression; also impairs plant growth and reproduction. Chemically induced mutants were created that reduce phytic acid but did not affect plant growth	Kim and Tai (2014); Jiang et al. (2019)
		Carotenoid biosynthesis cassette at two genomic sites produces high carotenoid content in the seeds and no detectable penalty in morphology or yield.	Dong et al. (2020)
	Soybean	Reduced levels of polyunsaturated fats by introducing mutations in two fatty acid desaturase 2 genes (FAD2-1A and FAD2-1B), which in the seed convert the monounsaturated fat, oleic acid, to the polyunsaturated fat, linoleic acid	Haun et al. (2014)
	Tomato	Gene edited tomatoes produce more vitamin D	Li et al. (2022)
		Incorporation of bacterial carotenoid gene encoding the enzyme phytoene desaturase increased β-carotene content by threefold, up to 45% of the total carotenoid content.	Römer et al. (2000)
		A ‘purple’ tomato, highly enriched with anthocyanins, was produced by the ectopic expression of two selected transcription factors from the ornamental flower snapdragon. These fruits also prolonged the life of cancer-susceptible mice.	Gonzali et al. (2009)
		Expression of a yeast S-adenosylmethionine decarboxylase gene (ySAMdc; Spe2) fused with a ripening-inducible E8 promoter to enhance phytonutrient content, juice quality, and vine life	Mehta et al. (2002)
	Wheat	Manipulation the expression of high molecular weight glutenin subunits – associated with elasticity of dough for bread making and cause of celiac in humans	Blechl and Anderson (1996); Alvarez et al. (2000)
		Two sgRNAs were used to target a conserved region adjacent to the coding sequence a-gliadin genes, leading to a strong reduction in a-gliadins – low gluten wheat	Sanchez-Leon et al. (2018)
Disease resistance	Various crops	Although technically useful solutions now exist for providing transgenic disease resistance, very few new GM crops with these traits have been introduced to the global market	Collinge et al. (2010); Collinge and Sarrocco (2022)
		Integrating antiviral genes of virus origin into plant genomes to aid resistance to viral pathogens	Niraula and Fondong (2021)
	Banana	Constitutive expression of apoptosis-inhibition-related genes in transgenic banana ‘Lady Finger’ showed high levels of resistance to Fusarium wilt	Chen et al. (2014)
		Increased resistance to banana bacterial wilt	Namanya et al. (2020)
		Improvement of B genome germplasm to produce hybrids of plantain (Musa balbisiana) leading to inactivation of endogenous banana streak virus (eBSV) in the B genome	Tripathi et al. (2019)
	Citrus (Citrus spp,)	Canker (Xanthomonas citri subsp. citri) resistant plants developed through editing the susceptibility gene promoter	Peng et al. (2017)
	Papaya	Control of Papaya ringspot virus disease	Azad et al. (2014)
	Peanut	Resistance to Phaeoisariopsis personata leaf spot disease	Vasavirama and Kirti (2012)
	Pearl millet (Pennisetum glaucum)	Expression of antifungal protein AFP from the mould Aspergillus giganteus provided protection against Puccinia substriata and Sclerospora graminicola	Girgi et al. (2006)
	Potato	Resistance to late blight, caused by the oomycete Phytophthora infestans which is only controlled by fungicides	Haverkort et al. (2009); Zhu et al. (2013)
	Rice	Constitutively express chitinase (chi11) and β−1,3-glucanase (GLU) genes improved sheath blight disease (Rhizoctonia solani) resistance	Kannan et al. (2017)
		Disruption of transcription activator-like effectors (TALEs) reduced expression of a family of putative sugar transporter genes conferring improved resistance to bacterial blight (Xanthomonas oryzae pv. oryzae)	Xu et al. (2019); Oliva et al. (2019)
		Editing at least one allele of the coilin gene considerably increased the resistance of the edited lines to infection with the potato virus Y and their tolerance to salt and osmotic stress.	Makhotenko et al. (2019)
	Tea (Camellia sinensis)	Resistance to blister blight disease (Exobasidium vexans)	Singh et al. (2018)
	Tomato	Trans-acting small interfering RNA control of geminiviruses – tomato leaf curl Gujarat virus	Singh et al. (2015)
	Wheat	Broad-spectrum resistance to powdery mildew (Blumeria graminis f. sp. tritici)	Wang et al. (2014); Zhang et al. (2017)
	Wheat and barley	synergistic efforts in traditional breeding and biotechnology should provide wheat and barley growers with Fusarium head blight resistant cultivars	Dahleen et al. (2001)
Processability	Banana	Downregulated ACC synthase or ACC oxidase can change banana fruit ripening process significantly through reduced ethylene production, decreased respiration rate, and delayed the change of peel colour in the ripening stage	Chen et al. (2014)
	Potato	RNA interference (RNAi) technology to silence the vacuolar invertase gene, which encodes a protein that breaks down sucrose to glucose and fructose; improves storage quality at low temperature	Clasen et al. (2016)
	Sugarcane (Saccharum officinarum)	Increased yields of directly fermentable glucose from lignocellulosic biomass for ethanol production	Jung et al. (2012)
Reproductive efficiency	Maize	I–CreI homing endonuclease from Chlamydomonas reinhardti to gene edit and provide male sterility for hybrid breeding systems	Djukanovic et al. (2013)
Phenotype	Banana	Used CRISPR/Cas9 to edit the MaGA20ox2 genes in ‘Gros Michel’ and obtained semi-dwarf mutants	Shao et al. (2020)
Lignin modification	Lucerne	Lignin in lucerne can be influenced using a transcription factor (Zhou et al. 2022).	Zhou et al. (2022)

There are numerous examples (Table 9) of other opportunities resulting from GM technologies that may benefit future crops for drought and salinity tolerance, nutritional and health benefits, disease resistance, improved processability, reproductive efficiency and plant phenotype. Cucumber (Cucumis sativus), a widely grown vegetable, can be impacted by various biotic and abiotic factors for which genetic engineering, which may include synthetic biology approaches within plants, has been proposed to provide plausible solutions (Bhatt et al. 2022; Lohani et al. 2022). Forages which do not receive research and development support to the same extent as row crops could be improved in terms of nutritional quality (Katoch 2022) and by using transgenes to reduce greenhouse gas emissions from ruminants (Winichayakul et al. 2020; Beechey-Gradwell et al. 2022; Roldan et al. 2022; Caradus et al. 2022b). Additionally, there has been a call for the development of genetically edited fruit crops to improve amongst other traits flavour, nutritive value, and superior phenotypes (Kanchiswamy et al. 2015). To date the only fruits marketed as GM is papaya – for resistance to a viral disease – papaya ringspot virus (Gonsalves et al. 2007), and the Artic^® apple developed to disrupt the expression of polyphenol oxidase enzyme and prevent unsightly browning in fruit flesh (Waltz 2015b; Stowe and Dhingra 2021). These second generation GM crops and forages have been slow to reach the market for two main reasons: continued consumer perceptions and technical and connectivity issues (Chan et al. 2020). Regulatory constraints mean that these new developments are often undertaken in controlled conditions and are therefore not tested as ‘tried and true’ in the field (Passioura 2020). Additionally, many of these GM-mediated traits are developed by public institutions with no ‘hard wired’ connection to international companies capable of taking them through national regulatory processes.

Concluding commentary

GM crops are one successful means of improving on farm productivity and profitability, the environment, and consumer benefits (Table 10). This does not exclude the use of other technology options for similar improvements. Indeed a survey of 208 on-farm developmental projects using low cost and locally available technologies and inputs in 52 developing countries, in which 8.98 million farmers have adopted these practices and technologies on 28.92 million hectares, representing 3.0% of the 960 million hectares of arable and permanent crops in Africa, Asia and Latin America, showed production increases of 37% per farm and 48% per hectare (Pretty et al. 2003). The practices and technologies used have led to increases in water use efficiency, improvements to soil health and fertility, and pest control with minimal or zero-pesticide use. This research reveals promising advances in the adoption of practices and technologies that are likely to be sustainable, with transparent benefits for the rural poor. However, neither of these approaches are mutually exclusive and GM technologies provide a viable option for some farmers (Raven 2014). The antagonism between approaches of agroecology and biotechnology toward delivering sustainable agricultural production (Heineman 2009) makes no sense and the fact that many consider them mutually exclusive defies logic and common-sense.

Table 10. Interpretation of evidence for benefits accruing from intended outcomes and risks associated with unintended outcomes from use of GM crops.

Consequence		Degree of benefit	Degree of concern	Comment
Intended	Economic gains	High in most circumstances, including small-holder farmers	Concern often associated with perceived or real influence of multinationals	Despite some concerns in developing countries
	Improved productivity and crop yield performance	High in most circumstances, including small-holder farmers	Lower increases in non-stressed environments	If correct, then should contribute to lower deforestation
	Crop management	Improved ease of management		Farmer uptake of GM crops has not been impeded by the need for on-farm management changes
	Crop product composition and quality	Good – generally compositional equivalence between GM and non-GM crops	Inferior lint quality in Bt cotton in Burkina Faso	Has been a fundamental component of human health safety assessment; use of Bt GM maize has reduced mycotoxins
	Agronomic and phenotypic plant traits	Generally equivalent to non-GM, even for stacked GM traits		Even when agronomic differences are documented they are unlikely to have biological or ecological significance
	Improved resistance to pests	High – achieving opportunities not available to non-GM breeding	A wider pool of insect control protein is required	Bt proteins generally not harmful to humans and target specific insect pests
	Reduced use of synthetic chemical	Generally a reduction through and Bt GM crops can be applied as part of Integrated Pest Management	Potential for producing resistance in weeds and insect pests	Negative environmental impacts reduced through using herbicide tolerant and pest resistant GM crops
	Lower C footprint and reduced greenhouse gas emissions	Some reduction through reduced use of petrochemicals	No data for reduction in nitrous oxide emissions	More than likely a positive outcome
	Reduced tillage	While tillage may have been reduced the environmental impact is unclear	Assumptions for reduced C lost from soil need tested	Some claims do not seem to be valid in practice

Consequence		Consideration	Degree of concern	Comment
Unintended	Increased herbicide tolerance of weeds	Developing herbicide resistance of weeds is not just restricted to GM crops	Can occur	Mitigated by use of herbicides with different modes of action
	Overuse of herbicides to mitigate tolerance of weed to glyphosate	GM herbicide tolerance continues to deliver significant economic and environmental gains	Can occur	Herbicide use seems to have increased on both GM and non-GM crops
	Overuse of pesticides to mitigate developing pest resistance to Bt proteins	Important pests have also become resistant to synthetic insecticides requiring increased use	Can occur	Bt crops may/can resolve some pest resistance issues to synthetic chemistry
	Secondary pests becoming dominant	If this does occur, then requires an to pest management systems	Rare, but some secondary pests can increase in importance	Secondary pests can be controlled by going back to insecticides, use of biological pest control, or use of multiple stacked Bt genes
	Impacts on non-target organisms	Often shown in laboratory studies but rarely in the field; some misleading evidence has been used to suggest non-target effects of Bt proteins	Bt proteins unlikely to have negative impacts on nontarget organisms	Bt proteins target a specific pest whereas crop protection chemicals may affect both intended target and beneficial organisms
	Resistant insect populations	Important pests have also become resistant to synthetic	Can occur	Mitigated by used of refuges and potentially use of a wider range of insecticidal proteins
	Increased weediness	GM arable crops are unlikely to survive for long outside cultivation	Unlikely
	Impacts on biodiversity	Preservation of biodiversity in natural ecosystems is of paramount importance	In GM crops biodiversity effects are low compared to use of synthetic chemistry	Any negative or less than desirable effect requires improvements in use are required through better management
	Transgene escape to wild populations	Many GM crop plants have wild relatives with which they can hybridise	Will occur in outcrossing species	Need to consider whether the transfer of transgene would be detrimental and create an unmanageable risk
	Impacts on rhizosphere microbes	Impacts depend on transformation events, experimental conditions and taxa analysed	Some effects but not necessarily deleterious	Synthetic chemistry can also potentially disrupt N cycling
	Horizontal gene flow	Poses negligible risks to human health or the environment	Unlikely and remote, requires effective selection processes	Gene transfer between bacteria in the phytosphere may be part of their evolutionary development
	Increased antibiotic resistance	GM plants most often uses an antibiotic-resistant gene as a selectable marker	Unlikely and remote, requires effective selection processes	No documented evidence of this occurring in 25 years of GM crop use
	Gene transfer to consumers	Gene transfer from GM food is no different from gene transfer in non-GM feeds	Highly unlikely
	Unintended compounds produced	Concern for unintended toxins being produced by GM crops has been proposed	No evidence of this	Bt GM maize may reduce mycotoxins in stored grain as a health benefit

Consequence		Consideration	Degree of concern	Comment
Other considerations	Safety of GM feed for animals	Testing for feed safety carried out on actual animal species	Equivalent to non-GM when fed to a range of farmed animals	Majority of GM crops are used for animal feed
	Food safety	No consistent and reliably documented issues; regulatory testing might that food from GM crops may be safer than that from non-GM crops	No human ailments, sickness or deaths linked to GM crops; food safety issues debates result from laboratory tests using rodents	Has caused polarised views; many non-GM crops also contain potentially toxic or allergenic compounds; level of risks to human health from foods contaminated with pathogens is far greater than from GM crops
	Nutritive value	Crops could be genetically transformed to improve their nutritive quality and value in an attempt to eradicate or lessen the impact of malnutrition	Nutritive value the same as non-GM or higher when selected for beneficial nutritive compounds	Uptake of nutritionally enhanced GM crops has been relatively slow for both animal and human consumption

This extensive review on the impacts of GM crops has shown that:

1. GM crops provide considerable benefits to farmers, consumers, and the environment;

2. GM technologies just like many non-GM technologies can bring risks, but these can be monitored and quantified and allow decision to be made about commercial, societal and environmental benefits versus real risks;

3. GM technologies are a valuable option that need to be promoted to solve current challenges and as a result improve not simply economic outcomes but also the environment;

4. Evaluation of outputs from GM technologies needs to continue to be de-risked before being made commercially available; and

5. While ‘checks’ and ‘balances’ are required, regulatory schemes need to focus on balancing risks and benefits and not just on ‘checks’. This is the situation currently for many countries including New Zealand were the HSNO Act 1996 needs review to allow a less adversarial path to the establishment of regulated field trials for research using containment to manage any risk.

GM crops provide food and feed that are already the most highly regulated biological technology in the world (DeFrancesco 2013; Baulcombe et al. 2014). There is a considerable body of good science relating GM technologies and their consequences, whether intended or unintended, and this must be the foundation for ensuring good and workable regulation. It has been considered that increases in regulation could generate further consumer distrust due to the misperception that risks are high (Herman et al. 2021). The desire to realise the potential of both older GM technologies and New Breeding Techniques (e.g. targeted gene editing) has been proposed to be more likely through transparent and open dialogue to inform regulatory systems (Mampuys and Brom 2015), but where this has occurred in the past it has not necessarily led to greater acceptance and uptake (Marris 2015). Recent reviews have been ‘pessimistic’ about the use of New Breeding Technologies (NBT) in future crop improvement endeavours (Lassoued et al. 2020).

New Breeding Technologies (NBTs) (synonymous with the term New Genomic Techniques (NGTs) (Parisi and Rodríguez-Cerezo 2021)) differ from older forms of transgenesis and genetic engineering in terms of precision and targeting of effects. NBTs include gene editing, targeted changes to a small number of bases of DNA using oligonucleotide-directed mutagenesis, cisgenesis, intragenesis, and the use epigenetic processes to change the activity of genes without changing a DNA sequence. So the question is should these NBTs be regulated differently? Currently in many jurisdictions, including New Zealand and the EU, NBTs are subject to the same regulations governing the use of genetically modified organisms (Zimny and Eriksson 2020; Purnhagen and Wesseler 2021; Caradus 2022). A revision of policies regulating GM crops is required with NBTs being the preferred future means for introducing new traits into crop and forage plants (Gould et al. 2022; Mbaya et al. 2022). Further the inclusion of NBTs innovations, at the very least, into organic farming systems would be a sensible and pragmatic option (Purnhagen et al. 2021).

Regulatory systems should be based on the benefit/ risk of the product not on the process/technology used to deliver the product (Smyth 2017b; Turnbull et al. 2021; Caradus 2022; Gould et al. 2022). Supporting that benefit/risk analysis a high level of trust is required in the organisations that evaluate and regulate GM crops (Siegrist 2000; Scott 2003; Ali et al. 2021), for society at large to accept the ruling and the use of GM technologies. To manage and understand potential risks associated with GM crops particular focus should include testing for:

1. Human and animal health and welfare impacts, including testing for allergenicity (EFSA GMO Panel et al. 2022a); and

2. Impacts on beneficial non-target organisms, principally arthropods (Romeis et al. 2008).

An awareness of gene flows from GM crops needs to be also considered and understood.

In New Zealand, there is an ever increasing list of foods from GM crops that can be sold (Food Standards Australia and New Zealand – FSANZ 2021), if appropriately labelled, but still the ability of farmers and growers to exploit the benefits of GM crops and forages is constrained (Caradus 2022). Adding to this inconsistency and lack of pragmatic logic is the fact that oil from 9 GM events in canola (Brassica napus), listed as approved on the FSANZ website, can be consumed as food by humans but the by-product meal, left after the oil extraction process, cannot be consumed as feed by animals in New Zealand. So the issue with foods from GM crops cannot be largely about safety to human health, which leaves then impacts on the environment and consumer acceptance of which the latter has been recently reviewed (Caradus 2022). So what are the environmental concerns that are stopping applications to use GM forages and crops in New Zealand? The Hazardous Substances and New Organisms (HSNO) Act 1996, administered by the Ministry for the Environment, which provides the regulatory oversight for release of GM organisms outlines minimum standards (Section 36) where the EPA can decline an application if a new organism is likely to:

1. cause any significant displacement of any native species within its natural habitat; or

2. cause any significant deterioration of natural habitats; or

3. cause any significant adverse effects on human health and safety; or

4. cause any significant adverse effect to New Zealand’s inherent genetic diversity; or

5. cause disease, be parasitic, or become a vector for human, animal, or plant disease, unless the purpose of that importation or release is to import or release an organism to cause disease, be a parasite, or a vector for disease.

And in Section 37 due regards must be given to:

1. the ability of the organism to establish an undesirable self-sustaining population; and

2. the ease with which the organism could be eradicated if it established an undesirable self-sustaining population.

The Environmental Protection Authority (EPA) is the organisation responsible in New Zealand for managing the effects of specified restricted activities on the environment which includes enforcing the HSNO Act. So while each application to EPA would be taken on a case-by-case basis, it is likely that all would be genetic modifications of non-indigenous forage and crop species (Zydenbos 2008), including species of Avena, Brassica, Hordeum, Lolium, Medicago, Pisum, Plantago, Trifolium, Triticum, and Zea. None of these are listed as unwanted organisms on the National Pest Plant Accord (NPPA) (MPI 2020), although Lolium perenne and some minor species (of no interest) of Trifolium and Plantago are listed on the consolidated list of ‘environmental weeds’ (Howell 2008). However, even so it is unlikely that GM variants would either affect New Zealand’s inherent genetic diversity or, based on extensive evidence from countries growing GM crops, cause disease. This leaves the issue of establishing self-sustaining populations and ease of eradication. In managed pastoral systems all forages used (with the exception of native tussocks (Caradus et al. 2022a)) are introduced and most are already established as self-sustaining populations along roadsides, bush clearings and in urban areas (Williams and West 2000). So would a GM variant of these forage species make this any worse or undesirable? Most forages are already conventionally non-GM bred for persistence and high yield, while any GM variant is most likely to be adding quality traits (Winichayakul et al. 2020; Beechey-Gradwell et al. 2022; Roldan et al. 2022; Caradus et al. 2022b). For GM crops (wheat, oats (Avena sativa), barley, and maize) these are generally annuals and therefore sown and harvested annually with few opportunities for self-sustaining population to establish.

Above all the provision of reliable and peer-reviewed information and commentary is required to provide confidence that risk tested GM crops can provide solutions and benefits to challenges facing the world with an ever increasing population. It is critical that there is responsible reporting of agricultural technologies to realise their potential (Mehta and Vanderschuren 2021). This may include publishing negative results where GM technologies may have failed – a call for more honesty in science (Mehta 2019), and for science to respect the need for the debate about the use of GM crops to include issues associated with freedom to operate, exploitation and societal benefit (Cayford 2004; Chetty and Viljoen 2007; Verma 2013). This is important because GM crops can offer the means to achieving stated aims in many countries of ‘zero hunger’ (Blesh et al. 2019) and ‘zero carbon’ (Vogt-Schilb and Hallegatte 2017; Fu-Chun 2021) resulting in improved human health and wellbeing, and an improved environment. However, acceptance of GM crops will be based on a belief resulting from impartial information provided from trusted sources. Having said that, upon reflection of the opposition mounted against GM technologies when earlier public controversy was caused by a deficit of publicly available information which then led to greater knowledge being provided and this then generally led to greater opposition (Levidow 2012). In effect, GM crops have been kept continuously on trial since their inception (Levidow and Carr 2007). There is an opinion that ‘it is likely that there will be a permanent difference in view of opinion that cannot be solved with more data or new facts’ (Mumpuys and Brom 2015). Their view is to ‘focus of the GM crop discussion should shift towards managing permanent different viewpoints and providing a platform for a broader conversation on agriculture and food production’. There is no doubt that fact-based conversations are required along with minds open to balancing risk and benefits rather than holding to ideologies and polarised views.

This systematic review on the benefits and risks of GM crops leads to the conclusion that GM crops provide considerable benefits and are a valuable option that needs to be employed to solve many of the current challenges facing mankind and as a result improve not simply economic outcomes but also the environment. GM technologies like many non-GM technologies can bring risks, but these can be monitored and quantified and allow decisions to be made about commercial, societal and environmental benefits versus real risks.

Disclosure statement

The author is employed by Grasslanz Technology Ltd which has an R&D investment portfolio that includes both genetic modification and gene editing of forages and microbes to provide mitigating solutions to current environmental and animal welfare issues facing both New Zealand and other pastoral economies.