Cumulative effects of fragmentation and development on highly productive land in New Zealand

ABSTRACT

Land fragmentation is a growing issue in New Zealand, however, no consistent or regular national monitoring has been established. A methodology for assessing land fragmentation was applied nationally for the first time, revealing that the greatest proportion of fragmentation occurred on land used for diffuse rural residence (>0.40 to ≤2.0 ha) and small parcels (>2.0 to ≤8.0 ha) with a 128% and 73% increase, respectively, between 2002 and 2019. In New Zealand, the most highly productive land (Land Use Capability (LUC) class 1, 2 and 3) is most impacted by continued fragmentation with 38%, 28% and 17% of baseline area, respectively, occupied by medium sized parcels or smaller (≤40.0 ha) with a dwelling in 2019. Impacts were greatest for Auckland with 40%, 44% and 25% of the region’s LUC 1, 2 and 3 land, respectively, occupied by small sized parcels or smaller with a dwelling, increasing to 64%, 67% and 47%, respectively, when including parcels ≤ 40.0 ha. Protection of LUC class 1 and 2 land, particularly, requires national attention. This metric provides an opportunity to evaluate land fragmentation and development over time that could serve both the assessment of policy performance and environmental reporting at national and regional levels.

KEYWORDS:

• Land fragmentation
• urban expansion
• versatile land
• cumulative effects
• New Zealand
• national policy
• national reporting

Introduction

Internationally, urban expansion and land fragmentation reduce the availability of agricultural land, food production and food security for local populations, as well as visual quality and greenspace of rural communities (Brabec and Smith 2002; Gardi et al. 2015; Nixon and Newman 2016). Loss of agricultural land has other consequences, such as increased pressure to convert natural or semi-natural land to agricultural use, such as in Africa, North America and Latin America (Harvey and Pilgrim 2011; Gardi et al. 2015).

Urban expansion and land fragmentation have been identified as a major threat to soil and potentially agricultural production in Europe (Gardi et al. 2015), Canada (Nixon and Newman 2016) and China (Song et al. 2015), often preferentially occupying high-quality soils (Zambon et al. 2018). Across 21 of the 27 European Union states, agricultural land area was estimated to have reduced by c. 70,000 ha per annum from 1990 to 2006 due to urban expansion, resulting in a loss of six million tonnes of wheat (Gardi et al. 2015). In Australia, urban sprawl is predicted to reduce local vegetable production from 82% to 21% of Greater Melbourne’s needs, where expected population growth by 2050 from 4 million to 7 million will require 60% more food (Sheridan et al. 2015).

In New Zealand, Māori have had a long connection and understanding of soil reaching back centuries to Polynesian migration, recognising soil as a taonga or treasured resource (Harmsworth 2020b; Harmsworth 2020a). Māori were the first in New Zealand to utilise what we now identify as highly productive land (Singleton 1988; Harmsworth 2020a) and were regarded as the first commercial orchardist and market gardener in the history of Auckland (Coleman 1967). In response to rapidly increasing European immigration in the eighteenth and nineteenth centuries extensive expansion in agriculture developed, as well as distribution of produce within New Zealand and later exports abroad. However, by the mid nineteenth century this began to change dramatically as the settler government enforced individualisation of land titles, and large areas of Māori land were confiscated, with a particular focus on prime agricultural land (Kingi 2008).

Today in New Zealand, it is widely recognised that highly productive land (Land Use Capability (LUC) classes 1–3), characterised as versatile for growing a range of crops, is a scarce and finite resource. National mapping of land use capability (LUC) estimates only 5.2% of New Zealand’s total land area has negligible or slight physical limitations for arable and horticultural use (described as LUC classes 1 and 2), increasing to 14.4% when including LUC class 3 land, defined as having moderate physical limitations (Lynn et al. 2009; Rutledge et al. 2010). This land is of particular importance in the New Zealand context for domestic food production requirements, which has previously been raised by rural industry (Horticulture New Zealand 2017). Relying on inter-regional and overseas supplies of basic food staples can create vulnerabilities in a societies’ food system. This is particularly apparent where there are increasing uncertainties surrounding the state of natural resources such as the availability of highly productive land, water scarcity and extreme weather events that can affect regional, national and global food production. The latter is particularly relevant with the current COVID-19 pandemic, which has highlighted the importance of food security and resilience with the disruption of food systems around the world (UNSCN 2020). Highly productive land is also important due to its contribution as a major export earner for New Zealand. Horticulture exports, in particular, have nearly doubled from $3.4 billion to $6.2 billion between 2010 and 2019 (Aitken and Warrington 2019). There is also growing global demand for dairy products, for which production is heavily reliant on highly productive land resources in New Zealand (Vogeler et al. 2014; Ministry for Primary Industries 2020b).

Pressures confronting highly productive land have been a long-standing issue dating back to the 1960s in New Zealand, with the historical location of cities often reflecting proximity to this finite resource for local food production needs (Hunt 1959; Coleman 1967; Cox 1968; Leamy 1975). In recent years, further concern has been raised on the potential effects of urban expansion and rural land fragmentation in the surrounding areas of some large cities, such as Auckland, Hamilton and Wellington (Rutledge et al. 2010; Mackay et al. 2011; Andrew and Dymond 2012; Curran-Cournane et al. 2016b; Curran-Cournane et al. 2018). Horticulture has been traditionally located close to labour supply and markets, which coupled with the lower costs of developing flat land and public concern about housing affordability, means horticultural land is typically more vulnerable to urban expansion than other rural sectors (Greenhalgh et al. 2017; Curran-Cournane et al. 2018). For example, from 2002 to 2016, New Zealand’s land area used for vegetables decreased 29%, from about 100,000 ha to about 70,000 ha (Ministry for Primary Industries & Ministry for Environment 2019). In Auckland, most of the contiguous highly productive land is located in and around Pukekohe and Franklin, and coupled with local climate, provides optimal growing conditions for vegetable production (Curran-Cournane et al. 2014; 2016a). Yet in 2019, it was calculated that legacy and current Auckland Unitary Plan zones have allowed various forms of development types to occupy at least 34%, 38% and 19% of LUC class 1, 2 and 3 land, respectively (Curran-Cournane 2019).

Internationally, several monitoring metrics have been developed, such as ‘landscape fragmentation’ due to transport infrastructure and sealed areas (Jaeger et al. 2008; European Environment Agency 2019), relative consumption of high-quality farmland (Song et al. 2015), and density of new urban areas (Hasse and Lathrop 2003). In New Zealand, there has been no nationally consistent, regular monitoring, evaluation or reporting on the status of highly productive land nor the land use pressures confronting this resource. While various regional efforts have been undertaken to develop metrics to monitor land fragmentation (Waikato Regional Council 2013; Curran-Cournane et al. 2014; 2016a; 2016b; Hart et al. 2013), methods and data consistency vary across regions. National one-off attempts have also been applied and approaches vary (Rutledge et al. 2010; Andrew and Dymond 2012), making it difficult to report on current state and trends (Ministry for the Environment & Statistics New Zealand 2018).

Recognising this long-standing issue, the Land Monitoring Forum (LMF), a Regional Council Special Interest Group formed by Local Government New Zealand, initiated development of guidelines to regularly monitor land fragmentation in a consistent way (Rutledge et al. 2015), with some preliminary application presented by Hill et al. (2018) and Jones et al. (2018). Land fragmentation in the current setting refers to any division of one or more aspects of a land resource available for land-based primary production, such as through subdivision and residential development including expansion of urban areas, that can restrict the current or future range of possible land uses (modified from Rutledge et al. 2015). At a generalised level, the objective of the guidelines was to focus on identifying land potentially available for land-based primary production and monitor the effect of ongoing fragmentation and development of land, particularly highly productive land. The term ‘development’ in the context of this study is all-encompassing considering both the expansion of urban areas as well as the occupancy of a dwelling on land parcels. While Rutledge et al. (2015) set out the rationale and method for establishing a land fragmentation indicator, it did not go as far as conducting the analysis. The objectives of this study were to refine this method and implement this work nationally, specifically to: (1) describe a robust and consistent methodology that can be applied at the national and regional level; and (2) for the first time, apply and report the results of land fragmentation and development analysis for a 17-year period spanning 2002–2019.

Methodology

The approach builds upon the methodology developed by Rutledge et al. (2015), to assess current and future land supply for land-based primary production. Improvements to the existing methodology included the use of several new geospatial data layers at different timesteps, quality assurance procedures, a parcel size classification, and nationwide implementation of the whole procedure. In addition to a national summary, three high growth regions – Auckland, Waikato and Canterbury – were explored for regional variation.

The data layer used to represent New Zealand’s cadastral land base was the New Zealand Primary Parcels – the Land Information New Zealand (LINZ) cadastral dataset – which is New Zealand’s official system to record and locate boundaries of land under various tenure systems (LINZ 2020b). For this study, datasets from the 2001/2002 period were compared to those of 2018/2019, nominally defined as the 2002 and 2019 timesteps based on the date of the NZ Primary Parcels layer. The 2002 reference timestep was chosen due to data availability – prior to this, linkages between land titles and cadastral survey registries were separately managed; they were brought together in the Landonline system in 2001 (LINZ 2020a).

A geospatial modelling analysis was performed using the ESRI ArcGIS platform and associated Python modelling code, in which the NZ Primary Parcels layer was filtered by a series of national geospatial data layers to identify areas of land that may have a restriction for primary production, as described in Table 1. These series of steps identified land-use restrictions arising from biophysical features (e.g. roads and water bodies), land ownership (e.g. legally protected natural areas), and urban areas. A key principle of this method was the use of open-source national scale datasets for each level of analysis. Benefits of public data include uniform, authoritative and independent characteristics, and the avoidance of issues including lack of access, inconsistent data, or restricted use (Rutledge et al. 2015). The other key principle was that datasets must have different temporal versions, with the likelihood of future versions for possible land fragmentation monitoring reiterations.

Table 1. Steps involved in geospatial analysis of LINZ Primary Parcels layer to identify land that may have a restriction for primary production.

Analysis step	GIS layer polygons attributes filtered	Source GIS layer
1. Removal of biophysical networks	Land Cover Database version5.0: Transport Infrastructure, River, Lake or Pond	LRIS (2020)
	LINZ NZ Primary Parcels: Hydro, Road, Railway, Streambed, Riverbed	LINZ (2020c)
	Topo50 NZ: Lakes, Ponds, Rivers, Canals, Lagoons, Reservoirs	LINZ (2020c)
2. Removal of urban land	Land Cover Database version5.0: Urban Parkland/Open Space, Built-up Area (settlement)	LRIS (2020)
	Topo50 NZ: Buildings, Residential Areas, Airports, Cemeteries, Golf Courses, Rifle ranges, Racetracks, Showgrounds, Sports fields	LINZ (2020c)
3. Removal of protected land	LINZ Protected Areas: Entirety of polygons	LINZ (2020c)
	QEII Trust Protected Areas: Entirety of polygons	QEII (2020)
4. Removal of parcels with a land-use limitation	Remaining LINZ NZ Primary Parcels without the ‘parcel_intent’ attribute of: DCDB, Fee Simple Title, Māori, Licence/Permit, Legalisation, Lease, Lease 20 years or more, and Statutory	LINZ (2020c)
	Land Use Capability: LUC class 8 Land Cover Database version5.0: Surface mine or Dump Topo50 NZ: Mines, Quarries, Gravel pits, Pumice pits, Landfills	LRIS (2020) LRIS (2020) LINZ (2020c)

Following the identification of land that may have a restriction for land-based primary production, the direct effects of land fragmentation and development were then assessed for the remaining land parcels by identifying changes in the definition (Table 2) and area of these land parcels through subdivision or changes in the parcel intent designation. Applying the methodology allows for the determination of monitoring changes in urban area, as well as fragmentation of the land supply for land-based primary production with the following metrics:

1. Changes in the summed area of parcels, and parcel counts, classified into specified parcel size class thresholds, with or without a dwelling (defined in Table 2).

2. Changes in parcel size class attributes associated with highly productive land (defined here as LUC classes 1–3).

Table 2. Parcel size class definitions.

Broad parcel size classes	Detailed parcel size
Diffuse urban residence	0.0 to ≤0.4 ha (with a dwelling)^1
Diffuse rural residence	>0.4 to ≤2.0 ha (with a dwelling)
Very small	0.0 to ≤2.0 ha (without a dwelling)
Small	>2.0 to ≤8.0 ha (with or without a dwelling)^2
Medium	>8.0 to ≤40.0 ha (with or without a dwelling)
Large	>40.0 ha (with or without a dwelling)

¹ Defined here as having one or more electoral address points.

² With or without a dwelling implies two components of the metric that are analysed separately.

The analysis of the above metrics were conducted separately for those parcels with or without a dwelling, indicating a place of residence identified by an electoral address point (Statistics New Zealand 2020a). There is no ‘off-the-shelf’ land parcel size classification used in New Zealand. This methodology adopted the classification (Table 2) proposed in previous research by the LMF, when looking at regional council implementation of the Rutledge et al. (2015) methodology. Diffuse urban and rural residential development were defined as parcels ≤ 2 ha with a dwelling located outside of the mapped urban areas. The 2 ha threshold was based on a noticeable ‘peak’ in the land parcel distribution, probably reflecting historical parcel classes used in planning, as well as commonly identified as ‘rural residential’ in the planning zones of a number of local authorities (Ministry for the Environment 2017). Analysis of Agribase^® 2020 in the development of this metric also showed that ‘lifestyle’ is the predominant land use associated with parcels of this size class. Diffuse urban residential development was identified as parcels ≤ 0.4 ha located outside of the mapped urban areas. This classification aligns with criteria for urban land used in the Overseas Investment Act (LINZ 2020c), and the Urban / Rural Classification (Statistics NZ).

The parcel size class boundaries are broadly based on historical parcel areas used in New Zealand, based on units of acres. For example, 2 ha is slightly smaller than 5 acres; 4 ha, 10 acres; 8 ha, 20 acres; and 40 ha is slightly smaller than 100 acres. The classes also broadly correlate to groupings of different land use recorded in the Agribase® dataset (AsureQuality 2020). Rural residential and horticultural land use tend to be predominantly in the small parcel size class. Unpublished analysis of Agribase^® 2020 showed a similar pattern to that found by Rutledge et al. (2015), when analysing Agribase^® 2014 data. The ‘broad acre’ farm types (dairy, sheep and beef, arable and forestry) were shown by Rutledge et al. (2015) to be predominantly classed in the medium to large parcel size classes as outlined in Table 2.

The development and land fragmentation of urban areas and parcel size classes were also analysed according to the Land Use Capability (LUC) classification in Lynn et al. (2009). The New Zealand Land Resource Inventory (NZLRI) LUC classification was produced and mapped across New Zealand from 1975 into the 1980s. Versatility is implied in the LUC classification system (i.e. it encompasses the soil, climatic, erodibility and topographic limitations or advantages of the land and soil attributes for a parcel of land). It treats class 1 land as the most versatile LUC class with versatility decreasing down the scale towards LUC class 8 land (Lynn et al. 2009). The focus is on LUC classes 1–3, which are proposed as the default basis for the mapping of highly productive land (HPL) in New Zealand through the National Policy Statement for Highly Productive Land (Ministry for Primary Industries & Ministry for Environment 2019). It is noted that prior to this there was no nationally consistent method for defining HPL, so care needs to be taken in comparing results with previous research using different methods. The fragmentation and development of LUC classes 1, 2 and 3 were also analysed individually because when considered collectively the dominance of LUC class 3 land can dilute effects on LUC classes 1 and 2 (Table A1).

One limitation of the LUC mapping was that areas that were already developed for urban use were not mapped for LUC, but rather denoted as ‘town’. Therefore the development of urban areas in 2002 and 2019 that are reported in this research are not the total urban area, but the ‘new’ urban areas that have occurred since the original LUC mapping, as the main focus is on reporting land fragmentation on LUC land between the two timesteps. When comparing change for specific LUC classes, the relative scale of these changes are also compared to baseline LUC areas for years 2002 and 2019, where the baseline LUC areas were defined as the land potentially available to support primary production, after the extraction of potential restrictions as defined in Table A1. The ‘new’ urban area extent as considered in this research (analytical step 2 in Table 1 and what coincided with mapped areas of LUC) was not excluded from this extraction process in order to determine how its expansion affects ‘baseline’ LUC areas across the two timesteps.

Results

Changes in development and parcel size classes 2002–2019

Expansion of ‘new’ urban area that occurred on land mapped as LUC classes 1–7, increased 29% (22,121 ha) from 76,755 ha in 2002 to 98,876 ha in 2019 (Table 3). As a result, the total New Zealand urban area increased 13% to 206,565 ha in 2019 (data not shown). In 2019, half of New Zealand’s urban area expansion was located in Auckland, Waikato and Canterbury (Table 3).

Table 3. Change in land area (with percentage change in parentheses) across urban and parcel size classes with a dwelling between 2002 and 2019.

		Land area (hectares)
	Year	New Zealand	Auckland	Waikato	Canterbury
Urban area^1,2	2002	76,755	10,624	13,025	14,358
	2019	98,876	14,406	15,693	19,043
	Change (%)	22,121 (29%)	3782 (36%)	2668 (21%)	4685 (33%)
Diffuse urban residence 0.0 to ≤0.40 ha	2002	6327	1077	792	740
	2019	10,753	1561	1553	1052
	Change (%)	4426 (70%)	484 (45%)	761 (96%)	312 (42%)
Diffuse rural residence >0.40 to ≤2.0 ha	2002	35,535	6509	4904	4142
	2019	80,982	13,582	13,232	7299
	Change (%)	45,447 (128%)	7073 (109%)	8328 (170%)	3157 (76%)
Small sized parcel classes >2.0 to ≤8.0 ha	2002	178,369	34,745	25,784	22,840
	2019	308,631	45,491	37,382	53,567
	Change (%)	130,262 (73%)	10,746 (31%)	11,598 (45%)	30,727 (135%)
Medium sized parcel classes >8.0 to ≤40.0 ha	2002	626,101	43,688	137,787	96,758
	2019	906,661	62,943	172,834	127,266
	Change (%)	280,560 (45%)	19,255 (44%)	35,047 (25%)	30,508 (32%)
Large sized parcel classes >40.0 ha	2002	3,936,921	68,569	590,598	753,761
	2019	5,770,097	97,078	806,358	1,140,207
	Change (%)	1,833,176 (47%)	28,509 (42%)	215,760 (37%)	386,446 (51%)

¹ with or without a dwelling is not applicable for urban area in this analysis.

² This is the ‘new’ urban area that has occurred on mapped LUC 1–7 land. ‘Town’ as featured by the NZLRI is available in Table A1.

Outside of the mapped urban area, the number of diffuse urban parcels (0.0 to ≤0.40 ha) increased 48% from 69,592 in 2002 to 103,212 in 2019 (Figure 1). The area of land in the diffuse urban residential category increased 70% (4,426 ha) nationally, and 96%, 45% and 42% in the Waikato, Auckland and Canterbury regions, respectively (Table 3).

Figure 1. Parcel type counts with a dwelling in 2002 and 2019 in New Zealand and select regions.

The greatest proportional change occurred in the area of land used for diffuse rural residential development (>0.40 to ≤2.0 ha) and small sized parcels (>2.0 to ≤8.0 ha) with a dwelling, which increased by 128% and 73%, respectively, between 2002 and 2019 (Table 3). The number of parcels in each category increased by a similar magnitude (Figure 1). The change differs between regions, with diffuse rural residential development in the Waikato increasing proportionally more than twice as much as in Canterbury, whilst Canterbury had an increase in small parcels that was proportionally over three times greater than in Auckland and Waikato (Tables 3 and 4 and Figure 2). Fragmentation and development predominately occurred in close surrounding of urban areas, satellite towns, and roading axes notwithstanding the extent of additional fragmentation beyond these zonal boundaries as illustrated in Figure 2.

Figure 2. Change in urban areas and parcel size classes, with or without a dwelling, for selected areas in Auckland (A and B), Waikato (C and D) and Canterbury (E and F) in 2002 and 2019. ‘Extracted land’ is that per analytical steps 1, 3 and 4 and ‘urban area’ is that per step 2 in Table 1. CBD denotes central business district.

Table 4. Change in land area (percentage in parentheses) of Land Use Capability (LUC) class 1–3 and 4–7 categories across urban and parcel size classes with a dwelling between 2002 and 2019.

		Land area (hectares)
		New Zealand		Auckland		Waikato		Canterbury
	Year	LUC 1–3	LUC 4–7	LUC 1–3	LUC 4–7	LUC 1–3	LUC 4–7	LUC 1–3	LUC 4–7
Urban area^1	2002	45,914	30,841	7790	2834	8503	4522	8918	5440
	2019	58,659	40,217	10,453	3953	10,276	5418	11,830	7213
	Change (%)	12,745 (28%)	9376 (30%)	2663 (34%)	1119 (39%)	1773 (21%)	896 (20%)	2912 (33%)	1773 (33%)
Diffuse urban residence 0.0 to ≤0.40 ha	2002	3271	3056	315	762	470	322	462	279
	2019	5533	5220	537	1024	911	643	638	414
	Change (%)	2262 (69%)	2164 (71%)	222 (70%)	262 (34%)	441 (94%)	321 (100%)	176 (38%)	135 (48%)
Diffuse rural residence >0.40 to ≤2.0 ha	2002	20,735	14,801	3038	3471	3003	1901	3192	951
	2019	43,252	37,729	5854	7727	8323	4909	5375	1923
	Change (%)	22,517 (109%)	22,928 (155%)	2816 (93%)	4256(123%)	5320 (177%)	3008 (158%)	2183 (68%)	972 (102%)
Small parcels >2.0 to ≤8.0 ha	2002	109,635	68,734	17,091	17,654	16,270	9514	17,957	4883
	2019	173,800	134,831	22,004	23,487	20,919	16,463	41,123	12,444
	Change (%)	64,165 (59%)	66,097 (96%)	4913 (29%)	5833 (33%)	4649 (29%)	6949 (73%)	23,166 (129%)	7561 (155%)
Medium parcels >8.0 to ≤40.0 ha	2002	384,378	241,723	18,048	25,640	87,906	49,881	72,240	24,516
	2019	494,498	412,164	26,055	36,438	102,512	70,322	87,321	39,945
	Change (%)	110,120 (29%)	170,441 (71%)	8007 (44%)	10,798 (43%)	14,606 (17%)	20,441 (41%)	15,081 (21%)	15,429 (63%)
Large parcels >40.0 ha	2002	1,063,194	2,873,728	14,135	54,434	186,654	403,944	278,714	475,048
	2019	1,390,135	4,379,962	22,581	74,497	237,369	568,989	353,686	786,522
	Change (%)	326,941 (31%)	1,506,234 (52%)	8446 (60%)	20,063 (37%)	50,715 (27%)	165,045 (41%)	74,972 (27%)	311,474 (66%)

¹ with or without a dwelling is not applicable for urban area in this analysis.

Between years 2002 and 2019, parcels with dwellings in the medium (>8.0 to ≤40.0 ha) and large (>40.0 ha) sized parcel classes increased in total area by 45% and 47%, respectively (Table 3), with the total number of parcels with dwellings increasing by 47% and 40%, respectively (Figure 1).

The observed increase in both the number of parcels and total area in all parcel size classes was a consistent trend of ongoing land fragmentation at both the national and selected regional levels. This is likely associated with multiple activities, including the fragmentation within parcels of larger size, as well as the development of land previously without a dwelling (Table 5). For example, the area of LUC class 1–7 land without a dwelling decreased by 26% from 11,463,389ha in 2002–8,523,426 ha in 2019 (Table 5).

Table 5. Change in land area (percentage in parentheses) of Land Use Capability (LUC) class 1–3 and 4–7 categories across parcel size classes without a dwelling between 2002 and 2019.

		Land area (hectares)
		New Zealand		Auckland		Waikato		Canterbury
	Year	LUC 1–3	LUC 4–7	LUC 1–3	LUC 4–7	LUC 1–3	LUC 4–7	LUC 1–3	LUC 4–7
Very small^1 parcels 0.0 to ≤2.0 ha	2002	49,592	49,333	2664	3412	6384	5130	5398	4160
	2019	36,240	35,792	1624	2463	4722	3708	3923	2871
	Change (%)	−13,352 (−27%)	−13,541 (−27%)	−1,040 (−39%)	−949 (−28%)	−1,662 (−26%)	−1,422 (−28%)	−1,475 (−27%)	−1,289 (−31%)
Small parcels >2.0 to ≤8.0 ha	2002	170,922	182,585	12,316	11,572	21,538	16,803	25,386	19,978
	2019	112,140	121,396	4911	5878	14,326	10,674	16,588	13,135
	Change (%)	−58,782 (−34%)	−61,189 (−34%)	−7,405 (−60%)	−5,694 (−49%)	−7,212 (−33%)	−6,129 (−36%)	−8,798 (−35%)	−6,843 (−34%)
Medium parcels >8.0 to ≤40.0 ha	2002	760,634	1,104,236	25,675	57,622	108,000	114,607	141,840	115,333
	2019	575,749	844,290	12,928	38,071	81,293	85,620	106,377	83,578
	Change (%)	−184,885 (−24%)	−259,946 (−24%)	−12,747 (−50%)	−19,551 (−34%)	−26,707 (−25%)	−28,987 (−25%)	−35,463 (−25%)	−31,755 (−28%)
Large parcels >40.0 ha	2002	1,005,752	8,140,335	15,280	104,825	112,219	720,120	243,358	1,516,685
	2019	715,877	6,081,942	7768	76,588	68,676	551,564	168,429	955,113
	Change (%)	−289,875 (−29%)	−2,058,393 (−25%)	−7512 (−49%)	−28,237 (−27%)	−43,543 (−39%)	−168,556 (−23%)	−74,929 (−31%)	−561,572 (−37%)

¹ The diffuse urban and rural residence categories in Table 4 are not used for parcels that do not have an associated dwelling on them, instead these are grouped together in Table 5 in the very small parcel category.

Overall, the national average non-urban parcel size with a dwelling decreased from 33.5 ha in 2002 to 28.0 ha in 2019, a 16% reduction, based on the area of land potentially available for primary production divided by the total parcel count (Table 3 and Figure 1).

Fragmentation and development of highly productive land and changes from 2002 to 2019

Except for large sized parcels, most of the fragmentation and development across all parcel sizes in New Zealand occurred on highly productive land (LUC classes 1–3) (Table 4). The rate of urban area expansion was also greatest on LUC 1–3 land, with c.25% more development since 2002 compared to the change on LUC 4–7 land (Table 4). In the Auckland and Waikato regions, the rate of expansion of the urban area onto LUC 1–3 land was at least double the rate of expansion on LUC 4–7 land (Table 4).

In 2019, more than half of the land area comprising the diffuse urban, rural residential development, small parcels (with a dwelling) and medium parcels (with a dwelling), occupied LUC 1–3 land in New Zealand (Table 4). There was an additional 28% and 25% decrease of LUC 1–3 and LUC 4–7 land without a dwelling between 2002 and 2019 (Table 5). Collectively, the trend suggests a large degree of fragmentation and development associated with all parcel classes during this 17-year period that disproportionally occurred on highly productive land. This is, for example, demonstrated by the growth of the small parcels in Canterbury, with an expansion of over 23,000 ha since 2002, with growth occurring three times greater on highly productive land (Table 4, Figure A1 and Figure 2).

When considering LUC class 1, 2 and 3 land individually, in 2019 all forms of fragmentation and development disproportionately coincided with LUC class 1 land followed by LUC 2 and 3 land in New Zealand. Compared to the national baseline areas (Table A1 – C), 15%, 10% and 6% of LUC 1, 2 and 3 land, respectively, was occupied by small sized parcels or smaller with a dwelling in 2019 (Figure 3 and Table A2). This increased to 38%, 28% and 17%, respectively, when medium sized parcels or smaller with a dwelling were considered (Figure 3 and Table A2). In contrast, only 5.2% of baseline LUC 4–7 land was occupied by medium sized parcel classes or less with a dwelling in 2019. Further fragmentation is also evident with an additional 724,129 ha or 21% of baseline LUC class 1–3 land (vs 8.3% of baseline LUC 4–7) occupied by medium sized parcels or smaller without a dwelling in 2019 (Table 5).

Figure 3. Cumulative effect of urban area and fragmentation of parcel size classes (with a dwelling) on baseline Land Use Capability (LUC) areas in 2002 and 2019.

The extent of change on highly productive land has also been substantial in New Zealand. Between 2002 and 2019 the area of land occupied by medium sized parcels or smaller with a dwelling increased by 20%, 32% and 47% on LUC 1, 2 and 3 land, respectively (Figure 3). A similar trend was observed at a regional level, but was most pronounced for Auckland with 40%, 44% and 25% of LUC 1, 2 and 3 land, respectively, occupied by small sized parcels or smaller with a dwelling in 2019, increasing to 64%, 67% and 47%, respectively, when medium sized parcels or smaller with a dwelling were considered (Figure 3). The latter reflects an increase of 100%, 38% and 44% on LUC 1, 2 and 3 land, respectively, since 2002. Only a small proportion (6%) of large sized parcels with a dwelling in Auckland occupy LUC 1 land in 2019. In contrast, large sized parcels occupied approximately one-third of LUC 1 land in Waikato (34%) and Canterbury (32%); although Canterbury experienced a large degree of development on LUC class 1 land between 2002 and 2019 (Figure 3).

Discussion

Implications of land fragmentation and development on potential primary production

The greatest proportional change in New Zealand occurred on land used for diffuse rural residential development and small parcels with a dwelling, with much of this occurring on highly productive land (Tables 3 and 4). Demand for lifestyle block living has been considered a key driver of land fragmentation, as are the financial gains associated with rural residential subdivision (Hart et al. 2013). While rural residential development is not necessarily viewed as a negative process in its own right, issues can arise with scattered, in-situ or sporadic development which can reinforce the demand for further fragmentation of rural land over time, therefore creating a cumulative effect (Hart et al. 2013). Over time, cumulative land fragmentation in rural production zones can subject commercial farming practices to increased reverse sensitivity pressures, which will not only impact negatively on operations but potentially on farm productivity (Curran-Cournane et al. 2016a; 2016b). Rural residential expansion can also be an issue where it coincides with highly productive land, with Andrew and Dymond (2012) previously estimating that lifestyle blocks occupied 10% of New Zealand’s LUC class 1 and 2 land, effectively shifting the land use on these sites out of commercial primary production.

New Zealand is a net food exporter (Rush and Obolonkin 2020) and is the world’s largest exporter of milk products (Ledgard et al. 2020), with growing demand as the global population continues to increase. The global population is projected to reach 10 billion by 2050 which will require sustenance of a healthy diverse diet that is sustainably grown (Willett et al. 2019). In New Zealand, population growth reached 5 million in 2020, taking only 17 years to reach from 4 million in 2003 – and is forecast to reach 6 million by the 2040s (Statistics New Zealand 2020b). Global and national literature repeatedly indicate that shifting the population towards more a sustainable diet is needed to ensure both health and climate benefits (Clark et al. 2019; Swinburn et al. 2019; Willett et al. 2019; Drew et al. 2020). Such shifts in diet are largely dependent on highly productive land, that food categories such as outdoor vegetable production, grains and a variety of fruit crops are limited to. Thus, the soil resource is still as important as ever, but at the same time is becoming increasingly vulnerable to land fragmentation and development (Figure 2 and Table A2). The global outbreak of COVID-19 has reinforced the importance of the reliable production of domestic basic food staples that highly productive land is pertinent to.

Recent targets have been set to add $44 billion in export earnings over the next decade by the primary sector to help with New Zealand’s economic recovery (Ministry for Primary Industries 2020a) – an ambitious target considering primary industries generated just over $46 billon in export revenue in 2019 (Ministry for Primary Industries 2020b). Increasing production in the primary sector will place greater importance on the availability, as well as the sustainable management, of highly productive land. To date, evidence suggests that the soil resource, in general, has not always been well managed, with soil quality issues such as soil compaction and the excessive use of fertiliser persisting for over 20 years in some regions (Taylor et al. 2017; Curran-Cournane 2020). It is likely that ongoing land fragmentation and development will see increased intensification on the remaining land resource, and consequential challenges in maintaining soil quality. Therefore, moves towards more sustainable land and soil use and management are incredibly important in achieving environmentally positive primary sector outcomes as pressures to increase productivity continue.

The results of the current study show that by 2019 one-third of New Zealand’s most versatile land (LUC 1 and 2) is now occupied by medium sized parcels or smaller parcels with a dwelling. This has implications in terms of options and choices for land-based primary production for current and future generations. For example, whilst recognising a large degree of variation can exist across rural sectors and regions, various forms of pastoral farming such as dairy, sheep and beef as well as certain outdoor vegetable and arable growing operations can require large areas of land to operate, ranging from >80 to several 100 ha (Statistics New Zealand 2016; Curran-Cournane 2017). Dairy farms typically operate on LUC class 1–4 land, with model simulations demonstrating greater operating farm profits generated from LUC class 1 and 2 land with profits declining on class 3 and 4 land (Vogeler et al. 2014; Vogeler et al. 2016). The latter is attributed to the generation of greater pasture production yields with less inputs required on higher LUC class land (Vogeler et al. 2014; Vogeler et al. 2016; Hicks and Curran-Cournane 2017). Research also consistently shows lower contaminant leaching on deep soils of LUC 1 and 2, compared to shallow and stony soils of lower capability, highlighting the importance of highly productive land to meeting freshwater quality objectives (Carrick et al. 2013; McLeod et al. 2014).

Implications for policy

Over the 2002–2019 period, land fragmentation and development disproportionately coincided with LUC class 1 land followed by classes 2 and 3 land. In response to the ongoing concern over the loss and fragmentation of highly productive land highlighted in the ‘Our Land’ national environmental report (Ministry for the Environment & Statistics New Zealand 2018), the Minister for the Environment announced in 2018 the instigation of work towards a proposed National Policy Statement for Highly Productive Land (NPS-HPL). A discussion document followed with an overarching objective to ensure ‘Highly productive land is protected for use in land-based primary production both now and for future generations’ (Ministry for Primary Industries & Ministry for Environment 2019). National policy statements (instruments issued under section 52(2) of the Resource Management Act 1991) enable government to prescribe objectives and policies for matters of national significance. The intention of the NPS-HPL is to ensure the protection of this remaining finite resource from effectively irreversible urban and rural residential expansion and fragmentation, as well as issues associated with reverse sensitivity.

Auckland, Waikato and Canterbury have Regional Policy Statement provisions that essentially, and at a generalised level, use the land use capability classification to instruct the protection of LUC 1 (Auckland); LUC 1 and 2 land (Canterbury); LUC 1, most subclasses of LUC 2, and some subclasses of LUC 3 land for Waikato; or LUC 2 and 3 where practicable in Auckland from inappropriate subdivision, use, and development (Auckland City 2008; Waikato Regional Council 2012; Environment Canterbury 2013; Auckland Council 2016; Waikato Regional Council 2018). Considering the impacts of fragmentation and development, particularly on LUC 1 and 2 land, the results of this study raise questions on the effectiveness of these policies on their own, notwithstanding other potentially influencing factors.

This reiterates the importance of the proposed overarching NPS-HPL, as well as measures to ensure its implementation and effectiveness, of which the latter could be evaluated against future updates of this land fragmentation metric. Future monitoring of this metric can also be specifically applied at the regional and territorial scale. For medium and high growth regions, identifying areas for future growth could be guided by future updates of this metric to avoid ongoing pressures of fragmentation and development of highly productive land. Regions that have already identified these ‘live zones’, for example, urban or countryside living zones that are not yet built on, could underestimate the amount of highly productive land impacted by land fragmentation and development in the current study.

When considering examples of international policy, British Columbia (BC), in Canada, is one of the oldest examples where land use regulation has legislated the Agricultural Land Reserve (ALR). The ALR, established in 1973 (Androkovich 2013), attempts to avoid land fragmentation and safeguard highly versatile farmland from development, which was previously identified as an ongoing issue for the region (Nixon and Newman 2016). Despite a 5% loss of original reserve area, the ALR, as a whole, has been regarded as being successful in maintaining local food production, sustaining the economic importance of the agricultural sector, and ensuring environmental protection (Androkovich et al. 2008; Nixon and Newman 2016).

While less focussed on highly productive land, and more on fragmentation as a whole, Inostroza et al. (2013) developed multiple metrics of spatial fragmentation monitoring over a 20-year period for South America that indicated an increasing trend towards urban sprawl (Inostroza et al. 2013). It signalled the need for an adequate urban development spatial strategy that can be achieved with the lowest cost and highest efficiency possible to accommodate for many cities forecast to double in size by 2035 (Inostroza et al. 2013). Not only was the information considered important in terms of ensuring the long-term availability of the finite land resource, it was proposed to be used to help inform what acceptable degrees of sprawl and fragmentation are.

Implications for future monitoring and reporting

The development of this metric is one that allows, for the first time, monitoring and reporting on the status of land fragmentation and development for all parcel classes, beyond previously published literature that has been largely focused on the expansion of urban areas and rural residential development (Rutledge et al. 2010; Andrew and Dymond 2012; Curran-Cournane et al. 2014). The impact of continued expansion of urban areas as well as diffuse urban and rural residential development onto LUC classes 1–3 is important, with definite differences in how this is occurring across different regions (Tables 4–5 and Figure 2). However, it is considered that the cumulative fragmentation effects of small and medium sized parcel classes also warrants the same, or arguably a greater level of future attention, particularly on LUC class 1 and 2 land as it impacts a larger proportion of this finite resource that can limit options and choices of the land for future generations.

The development of this land fragmentation metric is one that is reproducible, clearly and objectively defined, and based on publicly available data that collectively enables future regular monitoring and reporting. This fits criteria set by Statistics New Zealand for the provision of additional indicators needed for improving state of the environment reporting (Ministry for the Environment & Statistics New Zealand 2019), and allows filling a key knowledge gap identified by the Parliamentary Commissioner for the Environment (2019). Sound and objective definitions for describing diffuse residential development and a range of parcel size classes have been established that will allow consistent and comparable reporting over time. Some regions have their own specific land use capability classification systems that could be used as an alternative at the regional level (Agriculture New Zealand 1994; Hicks and Vujcich 2017), but because highly productive land is a national asset there is a rationale and need for national co-ordination of its definition, mapping and monitoring. Not only could this metric guide policy effectiveness monitoring and evaluation at national and regional levels, it could be established as a core indicator as part of the Environmental Reporting Act 2015 as well as other national reporting platforms such as Land Air Water Aotearoa.

Conclusion

A large amount of development on New Zealand’s highly productive land has been observed between 2002 and 2019, ranging from the growth of urban areas to diffuse urban and rural residential fragmentation. The greatest proportional change occurred for the latter, resulting in an area increase of 128% over the 17-year period. Most of this fragmentation and development occurred on highly productive land – a limited and finite resource.

However, the results of the study highlighted that highly productive land was most impacted by the fragmentation and development of small and medium parcel size classes, which arguably warrants greater attention. Nationally, this equates to 38%, 28% and 17% of LUC 1, 2 and 3 land, respectively, being occupied by medium sized parcels or smaller with a dwelling in 2019. A similar trend was observed at a regional level, but was most pronounced for Auckland with 40%, 44% and 25% of LUC 1, 2 and 3 land, respectively, occupied by small sized parcels or smaller with a dwelling in 2019, increasing to 64%, 67% and 47%, respectively, when including medium sized parcels.

Such trends are the result of either the fragmentation within parcels of larger size, as well as the development of existing land parcels that previously did not have a dwelling. This can create a cumulative effect and may reinforce the demand for further fragmentation that, over time, can compromise commercial rural operations that require large areas of land to operate, impacting options and choices for future generations.

Ongoing fragmentation and development is clearly disproportionately affecting the availability of New Zealand’s best farmland, particularly the highly versatile but scare LUC 1 and 2 land. Currently, there are no national policies that protect this finite resource from continued fragmentation and development. The trends shown in this research also raise questions on the effectiveness of current polices related to the protection of LUC 1, 2 and 3 land at the regional level. This metric enables consistent and regular reporting over time to evaluate continuous fragmentation and development, particularly of a scarce and finite soil resource, that could serve both the assessment of future policy performance and environmental reporting at national and regional levels.

Acknowledgements

The Ministry for the Environment and Statistics New Zealand are thanked for supporting this research. This research was also supported by Manaaki Whenua Landcare Research’s Strategic Science Investment Funding from the Ministry of Business, Innovation and Employment’s Science and Innovation Group. Waikato Regional Council and Olliver & Co are also thanked with supporting access to historical input datasets. The authors would like to thank the reviewers for their role which helped improve the quality of the manuscript.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by Ministry of Business, Innovation and Employment.

Appendix

Figure A1. Distribution of Land Use Capability classes for selected areas in (A) Auckland, (B) Waikato (C) and Canterbury as per Table A1 – A. CBD denotes central business district.

Table A1. National and regional baseline areas by Land Use Capability (LUC) class, pre and post selected modifications for years 2002 and 2019 (with proportions in parentheses).

Table

A. Unmodified baseline figures^1	Land use capability class (hectares)^4
	1	2	3	4	5	6	7	town
New Zealand	187,114 (0.7%)	1,201,446 (4.5%)	2,441,866 (9.2%)	2,775,064 (10.5%)	209,854 (0.8%)	7,467,891 (28.2%)	5,680,561 (21.4%)	145,727 (0.5%)
Auckland	4,383 (0.9%)	54,844 (11.2%)	64,473 (13.2%)	79,449 (16.2%)	0	172,642 (35.3%)	52,522 (10.7%)	44,342 (9.1%)
Waikato	46,333 (1.9%)	252,488 (10.3%)	279,767 (11.4%)	339,381 (13.8%)	10,305 (0.4%)	919,959 (37.5%)	395,838 (16.1%)	12,300 (0.5%)
Canterbury	23,119 (0.5%)	269,897 (6.0%)	544,570 (12.0%)	516,002 (11.4%)	24,275 (0.5%)	1,165,326 (25.8%)	712,135 (15.8%)	16,234 (0.4%)
B. 2002 modified baseline areas^2	Land use capability class (hectares)^4
	1	2	3	4	5	6	7	town
New Zealand	177,519 (1.0%)	1,138,627 (6.4%)	2,297,881 (12.9%)	2,460,749 (13.8%)	191,079 (1.1%)	6,350,304 (35.6%)	3,707,239 (20.8%)	116,945 (0.7%)
Auckland	4,127 (0.9%)	51,629 (11.7%)	60,534 (13.7%)	74,969 (16.9%)	0	160,169 (36.1%)	47,087 (10.6%)	36,698 (8.3%)
Waikato	44,462 (2.3%)	241,774 (12.5%)	264,711 (13.7%)	305,605 (15.8%)	9,286 (0.5%)	785,717 (40.7%)	226,137 (11.7%)	9,709 (0.5%)
Canterbury	21,908 (0.6%)	257,049 (7.2%)	518,508 (14.5%)	484,402 (13.5%)	21,844 (0.6%)	1,071,197 (30.0%)	589,829 (16.5%)	12,904 (0.4%)
C. 2019 modified baseline areas^3			Land use capability class (hectares)^4
	1	2	3	4	5	6	7	town
New Zealand	176,830 (1.1%)	1,136,796 (6.8%)	2,292,256 (13.8%)	2,440,475 (14.7%)	185,106 (1.1%)	6,176,926 (37.1%)	3,291,036 (19.8%)	116,315 (0.7%)
Auckland	4,101 (1.0%)	50,903 (11.9%)	59,713 (14%)	74,224 (17.4%)	0	156,481 (36.7%)	39,421 (9.2%)	36,426 (8.5%)
Waikato	44,229 (2.3%)	240,890 (12.5%)	264,209 (13.8%)	304,901 (15.9%)	8,998 (0.5%)	781,598 (40.7%)	222,812 (11.6%)	9,648 (0.5%)
Canterbury	21,754 (0.7%)	256,420 (8.6%)	517,116 (17.4%)	476,895 (16.1%)	18,062 (0.6%)	1,003,927 (33.8%)	404,272 (13.6%)	12,857 (0.4%)

¹ ‘Unmodified’ baseline LUC areas as reported by the New Zealand Land Resource Inventory. These figures are only supplied for contextual purposes and were not utilised in the analysis.

² ‘Modified’ baseline LUC areas after extracting biophysical networks, protected area features and other features such as gravel pits, landfills, mines, pumice pits, quarries (Table 1) as at either 2002 or 2019 timestep. ‘Urban area’ as defined in this research was not excluded from this extraction process in order to determine the effect of new urban expansion at that timestep.

³ As above for 2019.

⁴ Percentages do not round to 100% as land unsuitable for primary production (LUC class 8) and other LRI classes such water features and mines are excluded. Town figures are only provided for contextual purposes.

Table A2. Proportion of urban area and parcel size classes with a dwelling in 2019 by Land Use Capability (LUC) classes.

Table

		New Zealand		Auckland		Waikato		Canterbury
	LUC class	Hectares	% of baseline	Hectares	% of baseline	Hectares	% of baseline	Hectares	% of baseline
Urban area^1	1	5996	3.4%	438	10.7%	1756	4.0%	1068	4.9%
	2	24,995	2.2%	6715	13.2%	4448	1.8%	5373	2.1%
	3	27,668	1.2%	3301	5.5%	4073	1.5%	5390	1.0%
	1–3	58,659	1.6%	10,453	9.1%	10,276	1.9%	11,830	1.5%
	4–7	40,217	0.3%	3953	1.5%	5418	0.4%	7213	0.4%
Diffuse urban residence 0.0 to ≤0.40 ha	1	539	0.3%	25	0.6%	129	0.3%	51	0.2%
	2	2127	0.2%	288	0.6%	409	0.2%	269	0.1%
	3	2867	0.1%	225	0.4%	373	0.1%	318	0.1%
	1–3	5533	0.2%	537	0.5%	911	0.2%	638	0.1%
	4–7	5220	0.04%	1024	0.4%	643	0.0%	414	0.0%
Diffuse rural residence >0.40 to ≤2.0 ha	1	4622	2.6%	280	6.8%	1668	3.8%	321	1.5%
	2	16,641	1.5%	2770	5.4%	3638	1.5%	2199	0.9%
	3	21,989	1.0%	2805	4.7%	3018	1.1%	2855	0.6%
	1–3	43,252	1.2%	5854	5.1%	8323	1.5%	5375	0.7%
	4–7	37,729	0.3%	7727	2.9%	4909	0.4%	1923	0.1%
Small parcels >2.0 to ≤8.0 ha	1	16,502	9%	911	22%	3643	8%	2564	12%
	2	73,455	6%	12,537	25%	9518	4%	17,831	7%
	3	83,844	4%	8557	14%	7758	3%	20,728	4%
	1–3	173,800	5%	22,004	19%	20,919	4%	41,123	5%
	4–7	134,831	1%	23,487	9%	16,463	1%	12,444	1%
Medium parcels >8.0 to ≤40.0 ha	1	40,970	23%	967	24%	11,746	27%	4485	21%
	2	209,035	18%	11,672	23%	50,872	21%	34,802	14%
	3	244,493	11%	13,416	22%	39,894	15%	48,034	9%
	1–3	494,498	14%	26,055	23%	102,512	19%	87,321	11%
	4–7	412,164	3%	36,438	13%	70,322	5%	39,945	2%
Large parcels >40 ha	1	48,481	27%	235	6%	15,212	34%	6884	32%
	2	397,198	35%	7573	15%	100,951	42%	100,182	39%
	3	944,457	41%	14,773	25%	121,206	46%	246,620	48%
	1–3	1,390,135	39%	22,581	20%	237,369	43%	353,686	44%
	4–7	4,379,962	36%	74,497	28%	568,989	43%	786,522	41%

¹ with or without a dwelling is not applicable for urban area in this analysis.