Clever mothers balance time and effort in parental care: a study on free-ranging dogs

Mammalian offspring require parental care, at least in the form of nursing during their early development. While mothers need to invest considerable time and energy in ensuring the survival of their current offspring, they also need to optimize their investment in one batch of offspring in order to ensure future reproduction and hence lifetime reproductive success. Free-ranging dogs live in small social groups, mate promiscuously and lack the cooperative breeding biology of other group-living canids. They face high early-life mortality, which in turn reduces fitness benefits of the mother from a batch of pups. We carried out a field-based study on free-ranging dogs in India to understand the nature of maternal care. Our analysis reveals that mothers reduce investment in energy-intensive active care and increase passive care as the pups grow older, thereby keeping overall levels of care more or less constant over pup age. Using the patterns of mother–pup interactions, we define the different phases of maternal care behaviour.

TH Threat Individuals assume an aggressive posture with raised tail and ears, exposed jaws, fore legs stretched forward, producing a deep throated growl and staring at the recipient for a prolonged time.
VM Vomit Expels the semi digested and semi solid contents of stomach by regurgitation.

ESM 1.
A table showing the ethogram for maternal care. A unique two letter code was used for recording each behaviour during observations. The table shows the code, the name of the behaviour and its description. Maternal care was divided into active and passive care. Active care behaviours involved direct social interactions between the mother and pups. The passive care behaviours were not interactions, but individual actions of the mother occurring in the vicinity of the pups which allowed them to share time and space with and provide protection to the pups. In case of social interactions, the individual starting the interaction was designated as the initiator and the one towards which the behaviour was shown, was the recipient ( Individuals feeding on any solid or semi-solid food items. Individuals wander around and search thoroughly (by using the visual, olfactory or tactile cues) for food*.
Individuals clean their own bodies by licking, to remove dirt and parasites.
Individuals use their claws while sitting or standing, to scratch themselves. This is a self-maintenance behaviour.
Individuals pass their tongue over their body to make it clean.
Individuals sit, stand or lie down and remain immobile for several (at least 2) minutes at a stretch.

ESM 2
Details of the linear mixed effect model that shows the effect of pup age and their current litter size on the proportion of time spent in total care by the mother.
In order to check the effect of both the predictor variables i.e. pup age and mother's current litter size, we ran a "linear mixed effect model" incorporating the predictor variables as the "fixed effects" whereas the proportion of time spent by the mother in total care was included in the model as the "response variable". We collected the data on maternal care from 15 different dog groups that had 22 mother-litter units, over a span of 5 years (2010)(2011)(2012)(2013)(2014)(2015). Hence the identity of each mother-litter units (fgr) and the year of data collections (fyr) were incorporated in the model as the "random effects". A Gaussian distribution was considered for the response variable in the model. We started with the full model (mod1), i.e., with all possible two-way interactions among the fixed effects.

Variables used in the model:
Response variable: Proportion of time spent by the mother in total care-totpcprop Fixed effects: Age of pups in weeks-age

Current litter size-LS
Random effects:

Mother-litter unit's identity-fgr
Year of observation-fyr Details of the linear mixed effect model that shows the effect of pup age and their current litter size on the proportion of time spent in active care by the mother.
The predictor variables i.e. the "age" and "LS" were incorporated in the generalized linear mixed effect model to check their effect on the response variable i.e. the proportion of time spent in active care (acareprop) for 22 mother-litter units, collected over a span of 5 years. Identity of each motherlitter unit (fgr) and the year of data collection (fyr) were added in the model as the "random effects". We started with the full model (mod3), i.e., with all possible two-way interactions among the fixed effects.

Model validation
Since the residuals have an essential role in the model validation process, we did the "Bartlett test of homogeneity" of variances to check the presence of homoscedasticity in the model, separately for two predictor variables i.e. pup age (age) and current litter size (LS). P-value for the predictor "age" exhibited violation in the homogeneity assumption and it was fixed by adding "varFixed" as the "weight" in the model.

Variables used in the model:
Response variable: Proportion of time spent by the mother in active care-acareprop Fixed effects: Age of pups in weeks-age

Current litter size-LS
Random effects:

Mother-litter unit's identity -fgr
Year of observation-fyr Model: mod3<-lme(acareprop ~ age * LS, random= ~1|fgr/fyr, weights= varFixed(~age)) Here the two-way interaction between the age and LS showed significant effect on the response variable and hence we retained this model.

ESM 4
Details of the linear mixed effect model that shows the effect of pup age and their current litter size on the active care received per pup.
Total active care shown by the mother for a particular litter was divided equally among the pups that were observed to be present for the respective week of pup age. Active care received per pup was incorporated into the generalized linear mixed effect model as the response variable. We wanted to check the effect of predictor age and LS on the amount of active care received per pup. Identity of each mother-litter unit (fgr) and the year of data collection (fyr) were added in the model as the "random effects". We started with the full model (mod4), i.e., with all possible twoway interactions among the fixed effects.

Model validation
Since the residuals have an essential role in the model validation process, we did the "Bartlett test of homogeneity" of variances to check the presence of homoscedasticity in the model, separately for two predictor variables i.e. pup age (age) and current litter size (LS). P-value for the predictor "age" exhibited violation in the homogeneity assumption and it was fixed by adding "varFixed" as the "weight" in the model.

Variables used in the model:
Response variable: Active care received per pup-acareperpup Fixed effects: Age of pups in weeks-age

Random effects:
Mother-litter unit's identity-fgr Year of observation-fyr Model: mod4<-lme(acareperpup ~ age * LS, random= ~1|fgr/fyr, weights= varFixed(~age)) Here the two-way interaction between the age and LS showed significant effect on the response variable and hence we retained this model.

ESM 5
Details of the linear mixed effect model that shows the effect of pup age and their current litter size on the proportion of time spent in passive care by the mother.
The predictor variables i.e. the "age" and "LS" were incorporated in the generalized linear mixed effect model to check their effect on the response variable i.e. the proportion of time spent in passive care (pcareprop) for 22 mother-litter units, collected over a span of 5 years. Identity of each mother-litter unit (fgr) and the year of data collection (fyr) were added in the model as the "random effects". We started with the full model (mod5), i.e., with all possible two-way interactions among the fixed effects.

Variables used in the model:
Response variable: Proportion of time spent by the mother in passive care-pcareprop Fixed effects: Age of pups in weeks-age

Random effects:
Mother-litter unit's identity -fgr Year of observation-fyr Details of the linear mixed effect model that shows the effect of pup age and their current litter size on rate (frequency per hour) of care received (in terms of suckling and allogrooming) by individual pups.
The predictor variables i.e. the "age" and "LS" were incorporated in the generalized linear mixed effect model to check their effect on the response variable i.e. the rate of (frequency per hour) of care received (in terms of suckling and allogrooming) by individual pups (carercvd) for 22 motherlitter units, collected over a span of 5 years. Identity of each mother-litter unit (fgr) and the year of data collection (fyr) were added in the model as the "random effects". We started with the full model (mod7), i.e., with all possible two-way interactions among the fixed effects.

Variables used in the model:
Response variable: The rate of (frequency per hour) of care received (in terms of suckling and allogrooming) by individual pups -carercvd Fixed effects: Age of pups in weeks-age

Random effects:
Mother-litter unit's identity -fgr Year of observation-fyr