The Orchard data (orch) contain the results of the surveys of areas under certain species of fruit trees (apples, pears, peaches, apricots, oranges, lemons, small citrus fruits and since 2012 olives and vines intended for the production of table grapes). The statistical surveys on orchards are carried out every five years by the Member States. These surveys have been carried out since 1977.

The results presented in this database provide areas (in hectares) by variety, age and density classes by country and by production region. Data are mainly grouped in tables by fruit tree species. The following species are surveyed:

a) dessert apple trees,

b) dessert pear trees,

c) apricot trees,

d) dessert peach trees,

e) orange trees,

f) small-citrus fruit trees,

g) lemon trees and

h) olive trees.

The group small-citrus fruit trees (including tangerines and satsumas; clementines, wilkings and other similar citrus hybrids) is considered as a single species.
Data on plantations producing apples and pears as well as peaches on for uses other than dessert fruit were sent optionally by some countries since 1987.

The coverage of the species surveyed are presented in Figure 1:

In addition to the listed tree types, Bulgaria, Czechia, Germany, Greece, France Italy and Austria cover also other tree types listed in Table 2:

2017

12 countries declared they have a national legislation covering orchard statistics, while 11 declared that there is no such national legislation. Amongst those countries that have a legislation, Table 4 reports the year of entry into force, the eventual presence of differences between the EU and national legislation as well as the legal obligation for respondents to reply.

By observing the year of entry into force of the national legislation, Germany is the first one having applied such legislation in 1989 (amended in 2009 and 2014), followed by Denmark (2000), Romania and Croatia (2002, 2003), Bulgaria (2006), Belgium (2012) Greece, Italy, Lithuania, Hungary, Austria and Poland (2017). Only Croatian and Hungarian national legislation appear to have some differences with the EU legislation.

In all countries where a national legislation exists, there is obligation for respondents to reply.

In this section of the report, countries were asked to state whether the national crop item definition differ from the handbook ones, 20 countries answered “no”, France, Hungary and Slovenia answered “yes”. Differences stated by these countries are the following:

FR: Small citrus fruit trees in France 2017 = only mandarin, clementin

HU: Density and age classes are more detailed:

SI: In principal all fruits are produced for fresh consumption, so all areas are included in the dessert fruit areas. After the harvest, the product quality and market conditions determine the usage of fruits.

1.1.1.     Industrial processing fruits categories

All countries, which did not survey separately fruits for industrial uses, include them in the main categories (Table 5).

Countries provide lists of other fruit varieties not elsewhere specified (for dessert apple fruits, dessert pears, other oranges, other small citrus fruits, including hybrids). (Table 6).

TABLE 6 Other fruit varieties surveyed by country

The reference area covers the entire territory of the country; nevertheless, the following countries specified the further information:

EL: the special territory ofMount Athos is included (in NUTS 1: EL5-VOREIA ELLADA).

ES: Includes the overseas territories of Canary islands.

FR: Overseas departments are only surveyed for small citrus fruit trees.

CY: The data refer to agricultural activities only in the Government controlled area of the Republic of Cyprus.

PT: Mainland except Entre-Douro e Minho agrarian region. Autonomous Regions (Azores and Madeira) were excluded.

UK: Results are included for England as the areas of apples and pears grown in Scotland, Wales and Northern Ireland are very small, therefore no data collection takes place.

The entire territory of the country.

Orchard statistics are based on a total amount of 29 data sources in the EU, the most common ones are censuses and sample surveys (both at 38%), followed by administrative data (24%). None of the countries reported experts’ estimates or other data sources.

The amount and the type of source of data by country varies: 1.3 is the average number of sources by country in EU, most of the countries utilise one source of data, 5 countries utilise two sources of data and one uses three (Figure 3). A detailed list of data sources is in Annex 1.

Orchard statistics censuses are carried out in 11 of the 23 countries; in the structure of the Quality report they are listed 12 fruit tree types plus the category “other”). Based on such classification countries reported 39 different tree types collected with an average of 3.9 fruit tree types by country .

 “Dessert apple trees” is collected in 11 censuses, followed by “dessert pear trees” (9), “apricot trees” (5); “dessert peach and nectarine trees” are collected in 4 censuses while “Apple trees for industrial processing” and “pear trees for industrial processing” in 3. “Peach and nectarine trees for industrial processing (including group of Pavie)” and “table grape wines” concern 2 censuses, “small citrus fruit trees” 1 “other” 6 censuses (Figure 4).

The breakdown by country of tree type collected in the national censuses is illustrated in Figure 5. With 8 different tree types, Bulgaria has the widest coverage of orchards, followed by Germany, Hungary and Austria with 5 tree types, Czechia and Denmark with 4, the Netherlands with 3.

Countries indicate the lists used to build the census frame as well as any possible threshold used in the orchard statistics surveys. Ten countries provided this information illustrated in Table 7.

As pointed out in the table, thresholds are applied in nine countries and they span from a range of 0.01 ha in the Netherlands to 1ha in Lithuania.

Sample surveys are conducted in 11 countries; a wide range of fruit types are covered.

Most countries survey “dessert apple trees” (18% surveys), “dessert pear trees” (16%), “dessert peach and nectarine trees (13%), “apricot trees”, “small citrus trees” and “olive trees” are surveyed in 9% sample surveys. Other tree categories are lesser-represented (Figure 6).

Table 8 reports the thresholds applied in the sample surveys.

The lowest applied threshold is in Spain (0.01ha).

Figure shows the type of tree collected in the sample surveys, by country. Spain declared to survey “all” fruit tree types, Italy declared 10 different tree types, followed by Greece (9), France and Cyprus (8).

7 countries use administrative sources for orchards statistics.

The most common type of tree analysed in the administrative data sources are “dessert apple trees” (19%), followed by “dessert pear trees” (17%), “apricot trees” 14%, “dessert peach and nectarine trees” (11%). All other fruit tree types are below 10% (“other” tree types account for 11%).

Administrative sources of Slovenia and Croatia include the highest amount of tree type (10 and 9, respectively). All administrative tree type by country is illustrated below.

Three countries reported the main differences between administrative sources and the statistical definitions and concepts:

AT: IACS contains just the area information of fruit species without differentiation in variety and planting year; extensive fruit areas partly included.

SI: In register are included producers below production threshold.

SK: According to the national law (Act No 597/2006) there is an obligation for orchard users to register the orchard with the area at least 0,3 ha.

In the statistical survey (questionnaire "Osev") are collected the data on the orchards with area at least 0,15 ha.

Since 2009 the Statistical Office of the Slovak Republic does not collect the data on the orchards. The register of orchards has been established on the basis of results of the statistical surveys from the year 2008. The statistical data and data from the registered are compared twice a year (after sowing and harvest). Based on these comparisons, the Statistical Office of the Slovak Republic and ÚKSÚP exchange information on data quality.

Thresholds in administrative data are applied in Denmark, France, Slovenia and Slovakia.

none

With regards to the data collection methods in census sources, “Electronic questionnaire” is the most common (8), followed by “postal questionnaire” and “face-to-face interview”, both utilised in 5 countries; “telephone interview” is utilised in 4 cases. Denmark marked “other” methods, namely “paper questionnaires” (as an exception).

Figure 11 reports the data collection method by country. It is here interesting to notice that the majority of countries combine two different collection methods, Germany 3 and Czechia 4.

In sample surveys the most frequent data collection methods are “face-to-face interviews” (6) “postal” and “electronic questionnaires” (3). The other methods are employed in lesser cases (Figure 12).

Poland, ranks first in terms of data collection methods (3), followed by Latvia, the Netherlands and Romania with two. Other 6 countries reported just 1 collection method.

19 countries reported to proceed with both manual and automatic data validation, two countries (Bulgaria and Hungary) only indicated to use automatic validation; two countries (Slovakia and Sweden) only manual.

The data validation targets “completeness” (26%) and “consistency” (24%), 23% of them target “outliers”, 23% “aggregates” (20%). “Data flagging” is targeted in 4% cases; 3% indicate “other” targets[1].

Countries running the highest amount of data validation targets are Lithuania and Germany (5 targets), followed by 10 countries that run the validation against 4 targets (CZ, DK, EL, ES, CY, NL, AT, PL, PT, RO). The details of the data validation targets are reported in Figure 15.

not applicable

16 countries declared they have a quality management system in place for orchard statistics. Table 9 indicates the methods assuring the quality management, by respondent country.

TABLE 9 Quality assurance methods by country

Peer review is carried out in Greece, Austria, Romania and Slovakia (17% of countries). Table 10 reports the results of these reviews by country.

TABLE 10 Peer review carried out by country

19 countries reported to proceed with both manual and automatic data validation, two countries (Bulgaria and Hungary) only indicated to use automatic validation; two countries (Slovakia and Sweden) only manual.

The data validation targets “completeness” (26%) and “consistency” (24%), 23% of them target “outliers”, 23% “aggregates” (20%). “Data flagging” is targeted in 4% cases; 3% indicate “other” targets[1].

Countries running the highest amount of data validation targets are Lithuania and Germany (5 targets), followed by 10 countries that run the validation against 4 targets (CZ, DK, EL, ES, CY, NL, AT, PL, PT, RO). The details of the data validation targets are reported in Figure 15.

This indicator monitors the eventual quality developments since the last quality report by relevance, accuracy and reliability, timeliness and punctuality, comparability and coherence.

Figure 18 shows data for each of these indicators: apparently “timeliness and punctuality” showed highest improvements (in 6 countries), followed by “accuracy and reliability” and “Comparability” (improved in 4 countries).

“Coherence”   and “relevance” improved in 3 counties. As illustrated in Figure 18, there are cases in which the quality declined, namely the “accuracy and reliability” as well as “comparability” and “relevance”.  

The country-specific information on quality improvements is provided in Table 11.

More information on the quality management can be found in section 4 of the national quality reports.

Current orchard statistics Regulation EU No1337/2011 seems to meet relatively well the national needs as only Slovenia pointed out there are user needs unmet. Bulgaria, Latvia and Poland have however plans to implement improvements in order to meet better user needs. They are listed in Table 12.

Bulgaria, Greece, Spain, Lithuania and Austria run user satisfaction surveys for orchard statistics. The users appear to be “satisfied” (“highly satisfied” in Greece).

From the “additional comments” it is possible to conclude that although no official survey is conducted in Germany, the user needs are constantly monitored via periodical conferences among stakeholders. Also Bulgaria and Austria provided further details of their user satisfaction surveys which are reported in Table 13.

The data completeness rate is provided by all countries. In most of the cases the rate is reported to be 100% with some exceptions for Bulgaria, France Poland and Romania. The missing characteristics by country are reported below (Table14).

The overall accuracy indicator is a self-assessment provided by countries. 26% of countries assess the quality as “very good”, 61% “good” and 7% “medium” (Figure 19).

 “Coverage error” and “non-response error” are the main causes of a lower overall accuracy (both 26%), followed by “Measurement error” (19%), “Sampling error” and “other” had a lesser impact (16% and 13% respectively).

Figure 21 shows the main factors lowering the accuracy

Other comments are listed below:

DE: May be that in very few cases the farmers did not insert all fruit varieties of dessert apples/pears, but instead put them to the fruit trees for industrial processing because of less effort.

AT: Detailed information for special parameters not always available at the respondents.

PT: Survey results reflects the aging of Agricultural Sampling Base (BAA)

SK: At the time of the preparation of the file with orchards data we found in the Register small area of 12 ha, which was not removed from the Register.

11 countries provided information regarding the methods used to assess the sampling error in orchard statistics. 69% of these countries utilise the “relative standard error”, 16% “comparisons with the other data sources” and 15% utilise “other” methods. Figure 21 provides the detail of the methods utilised by country.

The coefficient of variation (in %) by tree type, is illustrated in Table 15.

The last row and the last column of the table report the average CV by fruit variety and by country, respectively. It could be noticed that on average the highest variation coefficient (in %) is the one recorded for “dessert peach and nectarine trees” (3.2% on average, mostly due to the high value of the coefficient in Poland; 11.8%). Poland, followed by Portugal, recorded the highest average variation coefficient (6.8% and 6.0%, respectively).

Please provide the coefficients of variation in %

In Czechia, Lithuania and Austria it appears that the sample frame includes wrongly classified units that are out of scope. The over coverage rate was 30% in Austria, 0% in Czechia, Germany, Lithuania and Sweden. The rest of the countries did not reply this question.

Sample surveys conducted in Greece, France, Italy, Cyprus, Poland and Portugal appear to include wrongly classified units that are out of scope.

Questionnaire is based on usual concepts for respondents in all countries that provide an answer, in both census and sample surveys.

Concerning the number of censuses performed with the current questionnaire (Table 17), the longest time frame is the one performed by the Netherlands for census (5) and by Greece for sample surveys (7).

Among the different measures taken in order to minimize the unit non-response rate in census, the most frequent ones are “reminders” (32%) “Follow-up interviews” (23%), “legal actions” (20%), “Imputation” (16%) and lastly, “weighting” (4%).

Actions taken by countries, in order to minimize the unit non-response rate, are indicated in Figure 24.

Unit non-response error is also analysed for sample surveys. The highest unit non-response rate is in the Netherlands (32%, judged as “moderate”) and the lowest one of Spain (1.25%, evaluated as “low”).

“Follow-up interviews” and “weighting” are the most common measures utilised to reduce the unit non-response rate of sample surveys (29%), followed by “reminders” (24%); “legal actions”, “imputations” and “other” measures score all 6%.

Figure 26 reports the country specific measures implemented.

Belgium, Czechia, Denmark, Germany, Greece, Spin, Croatia, Latvia, Lithuania, Hungry, Austria, Portugal, Romania, Slovenia and Sweden indicated that the data were released on the target day.

Italy answered “no” because of “Technical problems encountered and overlapping of other activities”.

An aspect covered in the Timeliness analysis is the reason of the possible long production times for orchards statistics. This question was answered by some countries in the following way:

BG: The final results are not published due to need of further analysis and control.

FR: Sample issues, with over coverage and under coverage that required treatments for re calculation of extrapolation coefficients.

HR: Workload, incompleteness of data in administrative source.

AT: Difficulties to get back all forms (questionnaires) in time.

UK: Resource issues and staff changes

not applicable

The analysis of the comparability over time is assessed by the length of comparable time series for orchards statistics; according to this indicator major breaks in the time series occurred in Czechia (in 2015), France (2016 = sample within FSS; 2012 = census changes in fruits surveyed) and in Hungary (in 2017 there were different methodology and threshold).

Coherence is monitored by the comparison of orchard statistics with other national data sources. They will be first presented the overall major cross-domain comparisons and then illustrated the country detail. 

Among the comparisons made by the countries as a whole, 41% of them is with “annual crops statistics”, followed by “FSS 2016 data” (29%) and “IACS” (16%), “other” kind of comparisons were made by 9% respondents and 4% made no comparison.

In Figure 29, the coherence cross-domain comparisons are shown by country. The Netherlands and Poland stated that they do not run cross-domain comparisons.

Table 21 reports the further comments countries made explaining the reasons of non-comparisons and the results of the comparisons.

TABLE 21 Coherence – cross-domain, reasons for non-comparison

Results should be expressed as percentage deviation from the corresponding areas in the orchard survey.

Accessibility and clarity are assessed by the dissemination formats (news releases, publications, online database), the documentation management and the quality of documentatio.

Figure 30 shows the dissemination formats available: 10 of the countries declared to publish news releases; 5 issue paper publications, 14 have electronic publications (6 among them, in English too), 11 countries have an on-line database.

Details on the accessibility by country are illustrated in Table 22.

9.2.          Accessibility – documentation on methodology

Another aspect of accessibility is the documentation on methodology. 10 countries have national reference metadata files on orchard statistics available; handbooks are available in 11 countries.

In total, 18 countries reported to have efficiency gains since the last data collection. Amongst the most common gains pointed out were “further automation” (31%), followed by “increased use of administrative data” (17%), “further training” (14%). “on-line surveys” and “other” gains scored 12% and 14% respectively (Figure 31).

With respect to the efficiency gains by country, Czechia and Austria reported 4, Poland 3.

Some countries pointed out other efficiency gains:

DK: Data on production of apples and pears were included in the survey for the benefit of crop statistics and collected together with the FSS survey.

DE: Improved questionnaire and automatic processing.

FR: "Orchard 2012 = census; Orchard 2017 = included in FSS 2016 sample".

IT: web monitoring of data and on line data entry.

PL: simplification.

21 countries experiment burden reductions since the previous quality report, Greece did not experiment any reduction, the United Kingdom did not reply.

The most common burden reduction measures in the countries as a whole, were “easier data transmission” (24%), “more user friendly questionnaires” (22%), “multiple use of the collected data” (16%). “Less variables surveyed”, “less respondents” and “other” burden reduction measures all scored 11%. (Figure 33).

Figure 34 reports the burden reduction measures adopted by countries since the last quality report. Latvia implemented 4 measures, Belgium, Italy, Hungary, the Netherlands, Austria, Poland and Sweden 3

Together with the set of aforementioned burden reduction measures, some countries pointed out other ones, which are here reported:

DK: Data on production of apples and pears were included in the survey for the benefit of crop statistics and collected together with the FSS survey.

ES: There are not personal interviews. The interviewers collect directly the data by direct observation

FR: Prefilled data with orchard 2012 census in order to reduce burden for respondents if no change since 2012

IT: The major efforts to reduce respondent concern:

       a) data collected by the survey were limited to the information strictly necessary to comply the Regulation;

       b) electronic questionnaire was used to facilitate the data collection;

       c) best estimates and approximations were accepted when exact details were not readily available;

d) burden on individual respondents was limited to the extent possible by minimizing the overlap with other surveys (FSS 2016).

HU: