Les migrations des Roms roumains en Europe Circulation within the Family Residential Area : The Families of Colombian Migrants in Spain

The residential geography of families of migrants is a topic rarely addressed from a quantitative point of view. Based on the analysis of survey data on Colombian migrants in Spain (ENI, 2007), this article aims at describing the network build up by the different places of residence of those family members, may these be in Spain, Colombia or anywhere else in the world. The links between the different places of residence are handled means of the circulation approach. Migrant’s circulation, but also monetary remittances and transnational communication by telephone or Internet, may help consider these family geographies as “family residential systems”. La géographie de l’espace résidentiel des familles des migrants est un sujet rarement abordé de façon quantitative. À partir de l’analyse de données d’enquête concernant les migrants colombiens installés en Espagne (ENI, 2007), nous décrivons les formes tissées par les différents lieux de résidence des membres de la famille, que ceux-ci se trouvent en Espagne, en Colombie ou ailleurs dans le monde. La mise en relation des différents lieux de résidence familiaux y est interrogée sous l’angle des circulations ; circulation des migrants eux-mêmes, flux monétaires et communication transnationale (par téléphone et par Internet), permettent de considérer ces géographies familiales comme de véritables « systèmes résidentiels familiaux ». La geografía residencial de las familias de los inmigrantes es un tema poco estudiado con métodos cuantitativos. A partir del análisis de datos de encuesta sobre inmigrantes colombianos en España (ENI, 2007), este artículo propone describir las redes dibujadas por los distintos lugares Circulation within the Family Residential Area: The Families of Colombian Mig... Revue européenne des migrations internationales, vol. 32 n°1 | 2016 26

The intersection between two research areas, "family" and "migration", merits analysis for several reasons.
Firstly, the family reveals information about individual migration.Indeed the migration of an individual should not be characterised as a simple transfer of his/her personal place of residence but should take account of the evolution of his/her entire living space.The conventional means of assessing this phenomenon is the census, emphasising the allocation of a migrant to a particular place, his/her place of residence.According to the census definition, migrants are persons who have transferred "their residence from a certain place of origin or point of departure to a certain destination or point of arrival" (Henry, 1981: 105).Contributions from geographers and demographers (Courgeau, 1975;Domenach and Picouet, 1995;Robette, 2012) have allowed migration to be perceived not simply as a transfer of a place of residence but as a change to a living space, i.e. change over time to one or several locations that make up this space 1 .The family residential area is an essential component of this living space; although these two areas are not superimposed, they often overlap.Thus, Dureau and Giroud (2014) have shown that the family's places of residence can structure the living space within which individuals circulate.
Secondly, migration can be conceived as a family reproduction strategy.Since Aristotle, the family has been considered as a unit of economic and social reproduction (in Politics I, Circulation within the Family Residential Area: The Families of Colombian Mig... Revue européenne des migrations internationales, vol.32 -n°1 | 2016 1253b).According to Engels (1976), building on Aristotle's theories, two functions coexist within the family unit: on the one hand work, enabling "production of the means of subsistence, of food, clothing, shelter and the tools requisite therefore"; on the other "the production of human beings themselves".In this Marxist tradition, some studies have considered migration as a reproduction strategy.In the case of international migration, Burawoy underlines the dissociation of the reproductive worker, i.e. the family, from the productive worker, i.e. the migrant (Burawoy, 1976); in Latin-American contexts, writers also understand rural-urban migration as a family reproduction strategy (Balán and Dandler, 1987;Dureau and Dupont, 1994;Quesnel and Del Rey, 2004).Migration has thus been perceived as a strategy to diversify economic activity.Such diversification allows economic uncertainties in the place of origin to be addressed.The multiplication of family locations corresponds to a development in the rationale of family reproduction, marking a shift from the "territorial community" to the "archipelago economy" (Quesnel, 2009).
The image of an archipelago evokes a family, whose members are scattered on islands structured around the common function of family reproduction.Studies on family dispersion which are not based on a statistical view of the family assigned to a household have therefore gradually been developed.Rather, these attempt to demonstrate the relevance of taking into account the family straddling several places of residence.In this context the notion of a residential system seems to be a fertile concept for thinking about the links between a set of places of family residence.This concept, which draws on African contexts (Le Bris et al., 1987), has been invoked in the analysis of South American (Dureau, 2004) and European (Pfirsch, 2009;Dureau and Giroud, 2014) contexts.The residential system is defined by a set of interacting places of residence.As long as access is maintained between the various hubs in the system of residence, dispersion of the family is not necessarily a barrier to reproduction, but represents a resource.The places of residence of the family thus form a system through interaction between these places.Osmont (1987), who formulated this concept with Le Bris, points to the movement of children, family events, transfer of the family name, family endogamy, material solidarity and economic exchange between members, as examples of interaction cementing the residential system in Dakar.However, while dispersion does not necessarily mean fragmentation, it cannot be said that dispersion inevitably entails the pooling of resources for the purposes of reproduction.As Ma Mung (1999) underlines, dispersion does not always mean continued interaction at a distance.In the context of international migration, the issue of the fragmentation of families is all the more relevant given the significant physical and legal 2 distance between places of residence of migrants' families.The forms of long distance interaction highlighted in the study of family transnationalism are therefore based on three major areas of research: financial remittances transferred from one place of family residence to another; movement of members of the family between places of residence; and regular remote communication between those members (Guarnizo, 2006).
Drawing on the example of Colombians in Spain, this article aims firstly to shed light on the specific features of the organisation of the family residential area of migrants, according to socio-demographic characteristics (sex, social category, year of arrival).The term "family residential system" is not generally used in the first part of this article, since we consider that its use is appropriate only where there is actual interaction between the various places of family residence, either through the movement of members or long-distance communication between members or regular transfer of financial remittances.In the case of international migration, dispersion does not necessarily entail transnational operation.These three forms of long-distance interaction within the family will be analysed in the second part of this article, which will examine the validity of the theory of non-fragmentation despite dispersion.
The families of international Colombian migrants residing in Spain are a suitable focus for such research.Firstly, Colombia is a departure country for a wide range of migration destinations, with Spain among the three most significant, together with Venezuela and the United States.Secondly, the Spanish National Statistics Institute (INE) conducted a quantitative study on migration networks in Spain in 2007, which gathered information on the places of residence of partners, children, parents, brothers and sisters of 1,029 Colombian migrants born in Colombia, representative of the 300,000 Colombians residing on Spanish territory at that time.Although anthropologists studying the Colombian family recognise the wide variety of family systems which can be observed in Colombia, partly due to the various Colombian settlement populations (European, indigenous and African), they agree on the primacy of the nuclear family with a very important role for the extended family in the organisation of daily life (Echeverri de Ferrufino, 2004).In the Colombian context, an approach defining the family as a group formed by the partner, children, parents as well as siblings, appears relevant.

Description of the Colombian Population Studied
The Encuesta Nacional de Inmigrantes (ENI) was undertaken in 2007 in the context of an unprecedented increase in the immigrant population in Spain and an absence of data allowing this contingent of migrants, who mainly arrived after the beginning of the 2000s, during the Spanish property market boom, to be studied (Pumares Fernández, García Coll and Asensio Hita, 2006).Studies on migration in Spain have underlined the "novelty, scale and speed" of change (Reher and Requena, 2009).In the space of a few years, Spain went from being a country of emigration to a land of immigration.Notwithstanding the range of migration flows which have affected Spain (Colectivo IOÉ, 2012;Sierra-Paycha, 2014) since the date of this survey (returns, multiple migration, new immigration), the ENI remains an unparalelled survey in Europe, due to the size of the sample (15,465 people of whom 1,029 were born in Colombia) and the detailed questionnaire used.This enables an in-depth study of the migration process in Spain on the eve of the great recession of 2008-2010.
A particular feature of Colombian migrants in Spain and Europe is the feminisation of migration flows.As illustrated by the population pyramid 3 (Figure 1), the modal individual among the Colombian population residing in Spain in 2007 is a woman of between thirty and thirty five years of age.In addition, the population is relatively young and of working age.The Colombian working population living in Spain in 2007 is a socially heterogenous group.Alongside managers and highly skilled technicians, there are unskilled workers (half of the women) and entrepreneurs.However, if the criterion of an upper secondary qualification (equivalent of A-levels or baccalaureate) is applied, the presence of a middle-class population can be observed.
The majority of Colombian migrants surveyed in the ENI (over 80%) arrived in Spain after 1999.Since 2002 and the end of Spain's favourable migration policy (no visa requirement for Colombian nationals), arrivals have slowed.Recent arrivals are under-represented in the survey and it is more difficult to select them in the INE sampling frames.
In order to describe the geography of family residence, it is necessary first to define what is meant by family.The ENI does not record all the places of residence of family members of the persons surveyed.While the places of residence of living children, parents and siblings are stated, the place of residence of partners is not adequately described.Indeed, although it is specified whether co-residents are partners (cónyuge o pareja in Spanish) without being formally married, it is unclear whether an individual has a partner living outside his/her accommodation.However, the place of residence of the spouse of the person surveyed is recorded, regardless of whether they cohabit.For this reason, the term 'partner' is used in this article, imprecisely, to refer to a spouse to whom the person surveyed is married.In the surveyed population (Figure 2), people without spouses generally represent the majority, regardless of age.It is only among individuals aged between thirty and thirty-nine years and between fifty-five and sixty-four that the percentage of married couples is over 50 percent.In reality, the proportion of those in unmarried relationships is greater.It should be noted that those who designed the questionnaire perceived marriage as the unique form of legitimate conjugality without cohabitation.According to the Colombian census, the lifetime fertility of Colombian women residing in Colombia in 2005 was 2.9 children per woman 5 (Gonzalez and Echeverri, 2009).In Spain, in 2007, the lifetime fertility of Colombian women was slightly lower, 2.27 children per woman, with variations depending on their level of education: 2.88 children per woman among the lowest skilled (primary level or no education), 1.91 among Colombian women with a secondary qualification, and 1.8 among those with higher education qualifications.However, the mothers of Colombian migrants have higher lifetime fertility, explaining the small proportion of only children in the sample (ninety-one out of 1,029) and the presence of very large families (with a maximum of twenty-one members).
If family composition is defined according to whether the subject has zero, one or two living parents, brothers and sisters, a spouse and children, there are twenty-four different forms of family structure.With regard to Colombian migrants residing in Spain, in most cases family units are families of "origin", i.e. with at least one living parent in 888 cases out of 1,029 and living brothers and sisters in 938 cases.There are fewer cases of families formed by the individuals themselves; only 455 Colombian migrants have a spouse but a majority (640) have at least one child.As a result, the modal family is a family of "origin": 164 migrants surveyed have two living parents, brothers and sisters but no partner or children.The second most common type of family (143 persons) is a "complete" family made up of two parents, at least one brother or sister, a spouse and at least one child.Where people have children, in the majority of cases they also have a spouse, however there are many with children but no partner.
The size of the family network, understood as the sum of the living spouse, children, brothers, sisters and parents, is variable (Figure 3); it ranges from one family connection to over twenty-one people.The distribution of those surveyed according to the size of their family reveals two local peaks, at five and eight persons., 1998).In our study (Figure 4), however, we observe that the size of the living family generally increases from aged sixteen to fifty-nine years, before falling slightly around aged sixty years.This development in family size during the life cycle also conceals an evolution in the composition of family connections.While it is dominated at the beginning of working life by the parental couple and siblings (Figure 5), the proportion occupied by children gradually replaces that of parents.The greater significance of the proportion of children among the oldest populations can be explained by a generational effect connected to an overall reduction in fertility indicators in Colombia at the end of the 20 th century (Dureau and Florez, 1996), implying lower lifetime fertility for the youngest migrants.the contrary a desire to create distance from the family".The way in which the family is spatially structured, which is neither necessarily a choice nor entirely a constraint, should be "interpreted as the 'best' possible arrangement to accommodate the specific operation of each family".
Family Residential Location Depending on the Nature of Family Connections and the Migrant's Age 18 A broad overview of the places of residence of families of Colombian migrants residing in Spain (Table 1) reveals a preliminary observation: only 24% of the family are co-residents, 30% live in the same Spanish municipe 6 or 37% if the scale is enlarged to take into account the whole of Spain.Thus, 63% of the family live in a country other than the hostcountry, (59% in the country of origin and 4% in a third country).The proportions of co-residents and residents in Spain vary depending on the nature of the link with the respondent: 86% of partners, 48% of children, 37% of fathers, 30% of mothers and only 4% of siblings are co-residents.The residential links with the country of origin, which we will refer to here as family attachment, are above all maintained as a result of the residence of parents and siblings, while the links to other countries of Colombian emigration seem to be generally ensured by siblings.Transnational configurations can therefore affect all types of connection but to a variable degree depending on the nature of the connection., 1998).The composition of co-residents also varies over the course of the life cycle, with siblings and parents in particular forming the largest proportion of co-residents from age sixteen to twenty years, before the presence of children and partners in the home replaces that of the original family members.At the end of working life, in the context of the Île-de-France region, children moving out leave the respondent living alone with the coresident partner (Golaz and Lelièvre, 2012).However, in the Colombian population studied (Figure 6), it is observed that the proportion of parents among co-residents hardly diminishes after twenty-five years of age, remaining above 20%, while the proportion of children is relatively high until the age of sixty, when it still represents around 10% of co-residents.Here we observe that family residential areas are very often spread over at least two countries while at the same time being characterised by a propensity to high levels of coresidence.

Family Residential Areas Illustrating the Diversity of Colombian Migrant Destinations
In order to understand family geography, by analysing the overall form rather than comparing the place of residence of each member individually to the place of residence of the respondent, we have developed a simple indicator to show the countries of residential dispersion of the family.According to this indicator of the structure of the network, family networks are either non-existent, where the respondent had no living family connection at the time of the survey; mono-polar, where all members live in the same country (even if this country is different from the country of residence of the respondent); bipolar (part of the family is in Spain, the other part in Colombia); or multipolar, spread across an area composed of Colombia, Spain and a third country.
The majority of family networks of Colombians residing in Spain in 2007 are bipolar networks (number of respondents = 637; 62%); multipolar networks form the second most numerous group (178; 17%), followed by Colombian mono-polar networks (128; 12%) and Spanish mono-polar networks (83; 8%).Finally, three respondents did not know where any of their living family relations were resident.
In a survey of 1,026 households in which the reference person stated that they had a family presence outside the household, 178 families were spread beyond the Colombia/ Spain pair.17% of families of Colombian migrants in Spain are therefore connected to other migration areas.These multipolar networks include several thousand connections spread across twenty-five countries.The distribution of countries of residence by the number of family connections highlights the importance of the "migration pair" in the residential structure of these multipolar networks (Colombia and Spain appear 709 and 510 times respectively as the place of residence of part of the family network of Colombians residing in Spain).The significance of this migration pair has a major impact on determining the distribution of places of residence, but a map of the countries of residence excluding the migration pair (Map 1) reveals the international spread of multipolar families.Beyond the migration pair, the main countries of residence of family relations are significant countries in Colombian migration history (123 family relations residing in the United States, fifty in Venezuela).Central American and European countries make up the other main locations of family of Colombian migrants in Spain.These results prompt us to move beyond the paradigm of the "migration pair" (Simon, 2002) to understand contemporary migration dynamics, of which Colombian migration is an example.A disaggregated approach to studying Colombian migrants in their various countries of emigration would appear to be a blind alley.Indeed, porosity between different areas of migration seems to be galvanised by family dynamics, revealed here in a transversal way through a "snapshot" of the geography of family residence of migrants.

Gender Impact on the Family Residential Area in the Migration Context
In general, studies on the proximity of the family show that gender has little effect on proximity, regardless of whether the context is French (Maison and Ortalda, 1998) or Colombian (Dureau, 2004).

Proximity of Partner and Children to Women Migrants
In this area, on the contrary, gender has a significant impact on the residential location of the family (Table 2).Women in couples more often cohabit with their partner (90%) and co-reside with their children (50%), than men (78% of whom live with their partner and 43% with their children).Beyond sharing accommodation, children and partners of women migrants are more often in Spain (93% of partners and 63% of children) than partners and children of male migrants (83% and 59% respectively).This leads us to the assumption that commitment to family life varies between men and women.Indeed, research on family transnationalism points to the dedication of migrant mothers to retaining proximity to their children and where that is not possible, to establishing strategies for circulating between the two countries of residence (Yépez et al., 2011).

Proximity of Siblings to Male Migrants
Parents live slightly closer to male migrants' place of residence than that of female migrants (43% of fathers and mothers of male migrants live in Spain compared to 42% and 40% respectively in the case of female migrants), but these differences seem minimal.Siblings tend to be closer to the place of residence of male migrants than to that of female migrants.In the case of male migrants, 6% of siblings co-reside, 17% live in the same municipe and 26% live in Spain, while in the case of women, only 3% of siblings co-reside, 9% live in the same municipe, and 19% live in Spain.

Siblings: Rooted in Colombia, Open to Other National Areas
The corollary of these different patterns of proximity of family residence, depending on the gender of the migrant and the nature of the family connection, is a more significant presence of partners and children of male migrants in Colombia (17% of partners and 38% of children of men live in Colombia compared to 7% of partners and 33% of children of women); and a more significant presence of parents and especially siblings of female migrants in Colombia (57% of fathers, 60% of mothers and 75% of siblings of women live in Colombia compared to 54% of fathers, 57% of mothers and 68% of siblings of men).
The family attachment to Colombia is therefore due to a greater extent to siblings than to any other family connection.According to Table 1, 64% of family connections residing in Colombia are made up of siblings of migrants despite the fact that these sibling relationships represent only 51% of family connections.Furthermore, more than half of parents live in Colombia.Siblings also ensure a connection with other national spaces outside the migration pair: 79% of family connections residing in a third country are sibling connections (although these only represent 51% of family connections).It is also among sibling connections that the proportion of residents in a third country is the highest (6% of siblings, 3% of children, 2% of fathers, live in a third country).In addition, the presence of a family member abroad is highly selective, since none of the migrants surveyed had a partner or a mother residing in a third country.
Significant disparities can be observed in terms of residential proximity depending on the gender of the migrant; while partners and children live closer to the residence of migrant women than to that of migrant men, parents and siblings however, live closer to migrant men's place of residence.In respect of both sexes, the attachment to Colombia is mainly through siblings and parents, and opening the family residential area to other national areas is highly selective and mainly exists through siblings and to a lesser extent children and fathers.

Differentiation of Family Residential Areas Based on Social Class
Traditionally, the upper classes have a wider area of family dispersion (Maison and Ortalda, 1998), thus occupying, as a family entity, a larger number of locations, in particular in the case of Colombia (Dureau, 2004).What happens in the context of migration?The size of the family is not the same depending on the social class (approached here on the basis of level of education, given that professions can be deceptive in the context of migration).When comparing higher skilled migrants (those who have completed higher education) to lower skilled migrants (those who have at the most completed primary school) 7 , the size of the family appears to be smaller in the case of higher skilled migrants (6.7 members on average) than in the case of lower skilled migrants (8 members).When migrants are compared on the basis of their level of education, neither age nor gender have any impact, since the median and mean, minimum and maximum ages and sex ratios are similar.

Highly Skilled Migrants Living near their Partners and Children
Migrants with higher education qualifications (Table 3) more often cohabit with their partners (77% of those who have completed primary education live with their partners compared to 92% of migrants who have completed higher education).Only 80% of partners of lower skilled migrants live in Spain, compared to 97% of partners of highly skilled migrants.The same rationale of residential proximity depending on level of education can be observed in relation to children.Few children live with lower skilled migrants (36%) but a majority live with highly skilled migrants (54%).Only 55% of children of lower skilled migrants live in Spain, compared to 76% of children of highly skilled migrants.However, parents reside much nearer to lower skilled migrants compared to highly skilled migrants (46% of fathers, 35% of mothers of lower skilled migrants live with them, while in the case of highly skilled migrants, the proportion drops respectively to 31% and 26%).More Significant Family Ties to Colombia in the Case of Lower Skilled Migrants 34 Family ties to the country of origin are greater for lower skilled migrants: 63% of the family resides in Colombia, compared to 59% in the case of highly skilled.The difference is small, however attachment is linked to different family connections depending on social class.In the case of the lower skilled, ties to Colombia are through partners and children (20% of partners live there compared to 3% in the case of highly skilled migrants; 43% of children live there compared to 17% in the case of highly skilled migrants).For highly skilled migrants, family attachment to Colombia is more often connected to parents.56% of their fathers and 63% of their mothers live there, while the proportion drops to 50% and 57% in the case of lower skilled migrants.Similarly, in the case of highly skilled migrants, parents form 27% of family connections living in Colombia, compared to 16% in the case of lower skilled migrants 8 .
Family Residential Areas of a more International Nature among Highly Skilled Migrants 35 Openness to migration to other national areas is more common in the case of highly skilled migrants (6%) than lower skilled migrants (3%).While in the case of lower skilled migrants, siblings are responsible for 83% of such migration, in the case of highly skilled migrants the role played by siblings diminishes (74%) in favour of children (17% compared to 12% of lower skilled migrants) and fathers (10% compared to 4% of lower skilled migrants) 9 .The children (8% versus 1%);the fathers (4% versus 2%); and siblings (8% versus 5%) of migrants who have completed higher education more often reside in a third country, than those of migrants who have only completed primary education.
Greater residential proximity to partners and children is therefore a characteristic of highly skilled migrants, while lower skilled migrants have greater residential proximity to their parents.Highly skilled migrants therefore engage in family migration.Family attachment to Colombia is more significant in the case of lower skilled migrants, while highly skilled migrants are more open to moving to other national areas; however these attachment and opening-up functions depend on class-based family connections.In the case of lower skilled migrants, ties to Colombia are more often through children and partners, while in the case of higher skilled migrants it is more often through parents.In contrast, migrating to other areas is more often due to siblings in the case of lower skilled migrants, while in the case of highly skilled migrants it is more often undertaken by the children and fathers.
One of the characteristics of family residential areas which emerge from this study is the class-based variation in transnational organisation.Lower skilled Colombian migrants appear to be more tied to Colombia (67% of family residential areas adopt a binational organisational pattern compared to only 54% of family residential areas in the case of highly skilled migrants), while highly skilled Colombian migrants more often have trinational family residential areas (26% of families of highly qualified migrants are spread over three countries, compared to only 14% of families of lower skilled migrants) 10 .This analysis demonstrates a contrast between the family migration of the upper classes within multipolar family setups, and the more isolated migration of the working classes, generating bipolar family structures.

The Family Residential Area as an Indicator of the Degree of Establishment
The pattern of residential location of the migrant's family has gendered and classed rationales; it also reflects the Colombian's year of arrival in Spain (Table 4).The proportion of the family living with the migrant increases slightly with the duration of the period of migration in Spain, ranging from 23% in the case of more recent migrants to 27% in the case of less recent migrants.However, when we look at the proportion of family connections residing in Spain, family proximity increases more significantly with the duration of the period of migration, from 32% in the case of more recent migrants to 50% in the case of less recent migrants.These trends are only significant in relation to the residential proximity of the partner and children.While only 61% of partners of more recent migrants reside with them and only 68% live in Spain, 94% of partners of less recent migrants live with them and 99% live in Spain.Similarly, while the proportion of co-resident children (24%) and children living in Spain (29%) is very low in the case of more recent migrants, 56% of children of less recent migrants live with them and 85% live in Spain.
The result of such trends is that the proportion of the family maintaining ties with Colombia gradually decreases as the duration of the migration period increases; in the case of more recent migrants, 62% of the family live in Colombia, while this proportion drops to 43% in the case of less recent migrants.The influence on the family residential area of the duration of migration shows how an approach centred on the family residential area complicates the temporality of migration.On the one hand, the residential location of the family acts as an indicator of the degree of establishment of the migrant, on the other hand, the location of the entire living space is relatively inert compared to the transfer of the place of residence.For example, even among less recent migrants it appears that a significant part of the family network (above 40%) remains rooted in Colombia.41 The degree of intra-family distance varies according to the social characteristics of migrants.It is greater in the case of recent migrants, lower skilled migrants and men isolated from their family network.In contrast, the distance is less significant in the case of highly skilled migrants, less recent migrants, and women who have greater means to migrate with their family, or to have their family nearby.Where the family lives at a great distance from the migrants, in almost all cases the place of residence is in Colombia, given that 94% of places of family residence outside Spain are in Colombia (Table 1).The descriptions of family residential areas set out in the first part of this article support the theory of significant family dispersion and lead us to explore, in the next part of the article, the flows connecting these various places of residence.
Long-distance Family Interaction: Communication, Remittances and Circulation 42 The aim of the third part of this article is to question the relevance of the theory of nonsegmentation, despite family distance, to analyse whether the migrant's new place of residence is connected to other places of residence within the family residential area, in particular, places of residence in Colombia.Studies conducted in the 1980s on family residential systems showed that the types of interaction generally used to describe a residential system were mainly mobility, communication and economic solidarity between the places forming such systems (Osmont, 1987).The field of migration studies incorporates these elements under the generic term of migratory circulation, i.e. all types of mobility generated by migration, including movement of goods, services, people, but also symbols (Doraï et al., 1998).We will therefore attempt to test the validity of each of these dimensions in order to understand whether these family residential areas are indeed family residential systems.Maintaining Long-distance Contact with Family in Colombia In the context of migration, it is evident that distance does not allow individuals to see each other physically on a daily basis.In order to maintain family links, migrants set up communication strategies using the standard means of communication available at the end of the first decade of the 2000s.
In our study (Table 5), the vast majority of Colombian migrants (97%) stated that they have repeated contact with family or friends still living in Colombia.Telephone is the preferred means of contacting family members in Colombia in 99% of cases and Internet (emails and chat) in 45% of cases.Indeed, communication depends almost exclusively on these two means since fewer than 10% of migrants use other means of communication (post, contact through an intermediary, etc.).According to the ENI surveys, out of all migrants in Spain, South-American groups maintain more frequent contact with the country of origin (Gaete Quezada and Rodríguez Sumaza, 2010).However, interaction is not on a daily basis.Only 16% of migrants phone their families in Colombia daily, and a quarter send emails.The pattern of such communication is usually weekly (76% of telephone communications, 71% of email exchanges), or monthly (99% of telephone communication and 96% of email communication takes place at least once a month).The use of modern means of communication to connect families living at a distance is therefore significant.It is likely that, in the past, without access to the Internet or telephone, segmentation would have been stronger and family connections would not have been maintained.Even in cases in which a migrant's entire family lives in Spain, the

Circulating between Spain and Colombia
However, physical meetings with other members of the family remain essential to maintaining family ties.Taking all periods of migration in Spain together, one-third of migrants have not returned to Colombia since their arrival.This ratio of non return to Colombia is relatively high and should be tested in relation to varying durations of periods of migration in Spain (Figure 7).Where such trips do take place, the main reason is to visit family members or friends (86% of reasons for trip).This reason may be combined with other reasons such as holidays (64%) or dealing with paperwork necessary for regularisation in Spain (8%).These visits are significant in terms of duration, particularly as the price of a plane ticket rises to nearly one thousand euros during the August holiday period.74% of stays in Colombia last at least one month.Maintaining family networks therefore represents a cost which has to be amortised by staying for a sufficient length of time.
Regardless of the type of family network which still has connections in Colombia (multipolar, Colombian mono-polar or bi-polar), between 66% and 70% of migrants have already returned to Colombia.This applies to the multipolar networks even though they are scattered across several countries, thus multiplying the costs of maintaining the network by the number of countries concerned.However, fewer than half of migrants with their entire family in Spain have returned to Colombia since their arrival in Spain, which shows the importance of family presence in Colombia to maintaining movement to and from Colombia.

Transfer of Financial Remittances
While communication appears to be a rather generalised phenomenon, since it is used by 97% of Colombian migrants, visits to the country of origin are less common (67%).What about the transfer of financial remittances as an indicator that strategies of economic reproduction are at work despite great distance?A study on migration channels based on the ENI shows that only 49% of migrants transfer remittances (Ibid.).The proportion of migrants sending regular remittances -65% -indicates a priori that only some migrants use mobility for the purposes of strategies of economic reproduction per se.
Where such transfers take place, they are usually sent on a monthly basis (68% transfer remittances at least once a month) rather than on a weekly basis (only 1% transfer remittances every week).While most transfer remittances at least once a year (86%), transfers may also sometimes be undertaken on an irregular basis depending on economic opportunities in relation to income: 14% of the Colombian migrants surveyed stated that they transfer money whenever they can.
Most Colombian migrants have an idea of the amount they transfer (79%).On average remittances amount to EUR 2,239 per year with variations depending on level of education (Table 6).The proportion of migrants transferring remittances broadly follows a U-shaped curve.While a larger proportion of lower skilled migrants send remittances (65%) compared to highly skilled migrants (51%), the largest group sending money are migrants who have completed upper secondary education (71%).Although highly skilled migrants are less likely to send money to Colombia, the mean, median and maximum amounts of annual remittances are greater among highly skilled migrants (EUR 2,669, 2,000 and 15,000 in the case of highly skilled migrants compared to EUR 2,027, 1,200 and 12,000 in the case of lower skilled migrants).However, this conclusion should be qualified since the group sending most money annually on average are migrants who have completed lower secondary education.51 When we consider the recipients of remittances transferred by Colombian migrants in Spain (Table 7), parents come at the top of the list of recipients (63% of migrants transfer money to their parents) followed by siblings (26%) and children (26%).As a consequence of the residential proximity of partners, the proportion of migrants transferring money to their partner is minimal (6%).Only 13.1% of migrants transfer remittances to recipients outside the family unit as defined in this article (partner, children, parents and siblings) which demonstrates the validity of the family perimeter, as defined above, with respect to strategies of economic reproduction within families.In this regard, it would be interesting to examine the frequency of transfers of remittances among the various transnational family connections, depending on their location.If we correlate transfers of remittances to the various family connections to the populations concerned by the presence of such ties in Colombia, we see that the hierarchy of connections with respect to family economic solidarity is different, since intergenerational links take precedence over collateral links.79% of migrants who have children residing in Colombia transfer money to their children in Colombia on a regular basis.76% of those who have parents in Colombia transfer money to them, while only 60% of migrants who have partners in Colombia transfer money to them on a regular basis and 31% of migrants with at least one sibling transfer money to them.Interpretation: Among the 667 migrants transferring money to Colombia on a regular basis, the partner is the recipient in only 5.7% of cases.However, where the migrant's partner lives in Colombia, the partner receives remittances in 59.1% of cases.
52 The proportion of persons producing flows between places of residence, irrespective of the content of such flows (flows of persons, money, or communication) represents a majority; such flows contribute to maintaining the family network, in particular with regard to the family members still living in the country of origin.However, the circulation of flows has a cost, both in terms of money and time, and we can assume that the intensity of circulation decreases as the duration of the migration period in Spain increases.

Does Circulation Decrease as the Duration of the Migration Period
Increases?
53 Above we found that the presence of family in Colombia tends to decrease as the duration of the migration period in Spain increases.Indeed, migrants who have been there the longest have more family in Spain than Colombia.Yet family is the main reason for travelling to Colombia, as well as being the main recipient of money transfers and the main contact in Colombia for transnational communications.Does this mean that flows between Spain and Colombia also diminish as the migration period in Spain increases, representing a sign of establishment? Figure 7 confirms such a decrease in relation to communication flows: while 99% of more recent migrants communicate with Colombia on a regular basis, the proportion of less recent migrants in communication is lower (90%).Cash flows evolve in a slightly different way with the proportion of migrants transferring remittances to Colombia increasing at the beginning of the migration process (51% of migrants who arrived between one and two years prior to the survey, 73% of migrants who arrived between five and six years prior to the survey) but dropping after six years spent in Spain (only 48% of migrants who have been in Spain for more than nine years transfer money to Colombia).The proportion of migrants visiting Colombia increases with the duration of the migration period; but as time passes, travel becomes less frequent.The mean period (in months) between the most recent stay in Colombia and the date of the survey increases with the duration of the migration period: the maximum mean period (fifty-two months i.e. more than four years) between their most recent stay in Colombia and the date of the survey applies to migrants who arrived in Spain nine years ago or more.Figure 7 shows a common break in the trends displayed by the three types of flows after seven or eight years spent in Spain.This result is related to the composition of the earliest cohort.This cohort is formed of individuals who for the most part arrived in the 1990s but some of whom arrived before the 1980s.The latter have minimal contact with Colombia.Their lack of involvement in flows between Spain and Colombia therefore causes discontinuity in the curves, exaggerating the trends observed.comprised of communications, mobility of migrants, and remittances, depending on the duration of migration.
56 As we observed in the second part of the article, the entire family is rarely located in close proximity to the migrant; only 8% of family networks are entirely situated in Spain.Attachment to Colombia is the main cause of intra-family distance, as members of the family are rarely present in a country other than the "migration pair", except in the case of highly skilled migrants.The study of long-distance family interaction of Colombian migrants in Spain is above all a study of family interaction between Spain and Colombia.This was the issue explored in the third part of the article.Long-distance family interaction through circulation, financial remittances or remote communication always represents the majority.Going against some studies which have pointed to the relative decline of the family as a socio-economic unit of reproduction (Theŕy, 2001;Weber, 2013), we observe that family blood relatives appear on the contrary to form connections based on economic solidarity, in particular in the case of intergenerational relationships (76% of migrants transfer money to their parents in Colombia and 79% to their children), but also in collateral relationships (31% of migrants transfer money to their siblings in Colombia).
While the rationale of strong remote interaction seems predominant, it is not universally shared.In a minority of cases, family segmentation (i.e.no connections with Colombia) is a fact.
57 Only 3% of migrants do not communicate regularly with Colombia, 35% do not transfer money on a regular basis and 33% have never returned to Colombia since their arrival in Spain.The periods of circulation to and from Colombia should however be considered in the context of the pattern of individual migratory journeys; indeed they seem to correspond to particular moments within the migration period in Spain, after several years of residency.Indeed, the concept of a 'family residential system' would seem to be a useful notion in studying individual Colombian migration.This concept enables us to qualify the originality of transnational family organisation; the transnational family is shown to be a specific case within the extended family residential system, in which at least two places of residence are separated by a national border.However with the exception of a difference in scale, we perceive no distinction in nature between the family residential system as developed in the context of internal migration, and its application to the transnational family.disappear and new locations appear; transplant where all places of attachment disappear in favour of the new places (Beltramone, 1975).

BIBLIOGRAPHY
2. Due to restrictive migration policies.
3. This pyramid differs from the one produced by Gonzalez and Echeverri, although it is based on the same data, due to their application of weighting and selection of a population born in Colombia, while we have opted to study a population which specified the department in Colombia in which they were born (Gonzalez and Echeverri, 2009).

4.
Unless otherwise specified, all tables, graphs and maps used in this article come from the 2007 ENI survey.

5.
Lifetime fertility refers to the average number of children born to a generation of women during their fertile years, without taking mortality into account.

6.
The municipe is the administrative entity below a province, which is the equivalent of a commune in France.

7.
Those with an intermediary level of education generally come between these two extremes in terms of family residential location.
8. These figures do not appear in any of the tables shown in this article.Percentages have been calculated from the rows in Table 3.

Ibid.
10.These figures do not appear in any of the tables shown in this article.They have been calculated according to a cross-table of migrants by level of education and form of family network.

ABSTRACTS
The residential geography of families of migrants is a topic rarely addressed from a quantitative point of view.Based on the analysis of survey data on Colombian migrants in Spain (ENI, 2007), this article aims at describing the network build up by the different places of residence of those family members, may these be in Spain, Colombia or anywhere else in the world.The links between the different places of residence are handled means of the circulation approach.
Migrant's circulation, but also monetary remittances and transnational communication by telephone or Internet, may help consider these family geographies as "family residential systems".

Figure 1 :
Figure 1: Population Pyramid of Colombian Migrants in Spain Circulation within the Family Residential Area: The Families of Colombian Mig... Revue européenne des migrations internationales, vol.32 -n°1 | 2016

Figure 2 :
Figure 2: Distribution of Colombian Migrants According to Marital Status by Age Group

Figure 3 :
Figure 3: Distribution of Colombian Migrants in Spain by Size of Family

Figure 4 :Figure 5 :
Figure 4: Mean and Median Sizes of Family Network by Age Group

Figure 6 :
Figure 6: Average Composition of the Family Network Co-residing with Colombian Migrants in Spain

Map 1 :
Countries of Residence of the Family of Colombian Migrants Residing in Spain (Excluding the Migration Pair) Scope: 1,029 Colombian migrants.Author: C. Sierra-Paycha, Philcarto, 2015.

54
These results highlight a change in the intensity of movement depending on the duration of the migration period, which recalls the concept of the 'mobile condition' developed by Imbert (2014).The "mobile condition" evolves in the course of an individual's journey; thus circulation practices can correspond to a specific period of time during the mobility trajectory.Similarly, in the case of Colombian migrants in Spain, circulation, for the most part intra-family, appears at its most intense when migrants are sufficiently established to set up circulation practices (transfer of remittances and return trips between Colombia and Spain), but also when there are still family members in Colombia with whom to exchange.

Figure 7 :
Figure 7: Proportion of Colombian Migrants in Spain "Maintaining Contact" with Colombia, by Duration of the Migration Period in Spain
Circulation within the Family Residential Area: The Families of Colombian Mig...

Table 1 : Breakdown of Places of Residence of Family Members of Colombian Migrants in Spain According to the Nature of the Family Connection (%)
Interpretation: In 2007, of the living brothers and sisters of Colombian migrants in Spain, 21% live in Spain, 73% live in Colombia and 6% live in a third country.
19 Of course residential proximity depends on numerous factors, particularly age.In France the proximity of the family increases with age (Maison and Ortalda Circulation within the Family Residential Area: The Families of Colombian Mig...

Table 2 : Breakdown of Places of Residence of Members of the Family of Colombian Migrants in Spain by Sex and Nature of Family Connection (%)
Circulation within the Family Residential Area: The Families of Colombian Mig... Revue européenne des migrations internationales, vol.32 -n°1 | 2016Interpretation: Among living fathers of Colombian migrants in 2007, 37% live with these migrants, 42% live in Spain, 56.5% in Colombia et 1.5% in a third country.

Table 3 : Breakdown of Places of Residence of Members of the Family of Colombian Migrants in Spain by Level of Education and Nature of Family Connection (%)
Circulation within the Family Residential Area: The Families of Colombian Mig...
Revue européenne des migrations internationales, vol.32 -n°1 | 2016Interpretation: As of 2007, among the partners of Colombian migrants without education or who have only completed primary education, 77% live with these migrants, 80% live in Spain, and 20% live in Colombia.However, 92% of partners of migrants who have completed higher education live with these migrants, 97% of these partners live in Spain, and only 3% live in Colombia.
Circulation within the Family Residential Area: The Families of Colombian Mig...

Table 4 : Breakdown of Places of Residence of Members of the Family of Colombian Migrants in Spain by Nature of Family Connection and Year of Arrival (%)
Circulation within the Family Residential Area: The Families of Colombian Mig... Interpretation: Among the ninety-two Colombian migrants who arrived most recently in Spain, i.e. less than two years prior to the survey, only thirty-one had a partner at the time of the survey.While 61.3% of those partners lived with the migrant, 32.3% still lived in Colombia.

Table 5 : Modes and Frequency of Communication between Colombian Migrants in Spain and their Family or Close Relatives in Colombia
Among the 994 Colombian migrants in regular contact with family or close relatives in Colombia, 99.8% use the phone and 45.2% use emails and internet chat.16.1% of migrants who contact their close relatives by phone do it on a daily basis, and 75.5% at least once a week.Since several responses were possible, with migrants able to communicate by phone as well as by mail, the total exceeds 100%.
Circulation within the Family Residential Area: The Families of Colombian Mig... Revue européenne des migrations internationales, vol.32 -n°1 | 2016 level of communication with Colombia is high (85%), but lower than in other configurations (97% of Colombian bipolar networks, 98% of multipolar networks and 100% of mono-polar networks communicate on a regular basis).

Table 6 : Amounts of Remittances Transferred by Level of Education (€)
Circulation within the Family Residential Area: The Families of Colombian Mig...

Table 7 : Transfer of Remittances by Type of Recipient (%)
Circulation within the Family Residential Area: The Families of Colombian Mig...